The Missing Link: How Languages Can Influence Conflict

A Thesis in the Field of International Relations

for the Degree of Master of Liberal Arts in Extension Studies

Harvard University

November 2022

Copyright 2022 Megan Ashley Kregel

Abstract

The purpose of this research project is to ascertain whether there is a link between the languages of the world and the conflicts that occur yearly. Can the number of languages present in a country be a contributing factor to conflicts arising there? Is there in fact a correlation between the two? Given the limited amount of prior research on this topic, this preliminary research, conducted through use of geographic and conflict data. has been exploratory, and therefore results may vary at given times for different conflict years.

Author’s Biographical Sketch

Megan Ashley Kregel completed her undergraduate study at Indiana University-Bloomington, where she graduated in 2011 with degrees in History, Eastern European Languages and Literatures, and Eastern European Studies. During the last eleven years she has worked in the localization industry specializing in rare language recruitment, medical/pharmaceutical translations, and training initiatives. Her background and several Harvard Extension School courses led her to investigate if there is a link between language and conflict.

Table of Contents

Frontispiece............................................................................................................ iv

Author’s Biographical Sketch................................................................................. v

Dedication............................................................................................................... vi

Acknowledgments................................................................................................. vii

Chapter I Introduction............................................................................................. 1

Chapter II Background of the Problem................................................................... 3

Linguistic Background.................................................................... 3

Conflict Research Background........................................................ 5

Political Background....................................................................... 7

Geographic Background.................................................................. 9

Chapter III. Definitions......................................................................................... 12

Chapter IV.? Research Methodology..................................................................... 16

Chapter V.? Research Limitations? Known Pre-Research Limitations.................. 21

Post-Research Limitations Acknowledged........................................................... 21

Chapter VI.? Results.............................................................................................. 24

UCDP/PRIO Armed Conflict............................................ 24

UCDP Dyadic Dataset....................................................... 30

UCDP One-Sided Violence Dataset.................................. 36

UCDP Non-State Conflict................................................. 41

UCDP Battle Related Deaths............................................. 45

Deadly Electoral Conflict.................................................. 50

Regressions........................................................................ 54

Chapter VII. Conclusion........................................................................................ 58

References............................................................................................................. 60

Chapter I Introduction

The languages we speak help us to understand the world around us. They not only shape ?the nature of our society, but impact us psychologically as well. Our “communication is based on features that describe an event and capture emotions, needs, interests and fears. Language is used to resolve or escalate dispute.”[1] ?But could language in fact be one of the reasons that a conflict occurs within a particular region at all?

While linguists have acknowledged that language could be one of the main causes of conflict, in addition to social, economic, or political factors, scholars in other fields have not fully considered or dismissed the linkage between them. I hypothesize that if there are more than ten languages spoken in any given country, the probability that a conflict will occur is greater than if residents speak fewer languages, given that communication will likely be far more difficult in those countries where the people speak more languages. For countries that have fewer languages, individuals can typically communicate more effectively, thus establishing a peaceful status quo. Language, in effect, could contribute to peace and democracy.

Due to the limited number of resources and literature on this assertion, the following chapters will define what information there is on this topic, and additional chapters will detail the methodology used in testing.? The final chapters will contain research findings and statistical outputs from exploratory regressions.

?

Chapter II Background of the Problem

Linguistic Background

?

?

Language is all around us.? It is not only our means of daily communication, but also acts as a part of our ethnic identity.? Therein lies a complication and topic of great debate amongst scholars. “Language” is often used interchangeably with “ethnicity,” which Christopher Anderson and Aida Paskeviciute argue is not the case.? They state that:

“[e]mpirically speaking, researchers have commonly used language as a proxy for ethnic variation in a country and have utilized so-called measures of ‘ethnolinguistic fractionalization’ to capture variability within a population along ethnic lines. […] Suffice it to say for our purposes that using language as a proxy for ethnicity or relying exclusively on ethnicity or language to measure population heterogeneity is likely to miss important dimensions of variation within a population for two reasons: First, it assumes that ethnicity maps onto language one-for-one; second, it assumes that language has the same consequences for citizenship behaviors as ethnicity.”[2]

?

Interestingly, from a linguistic perspective, it is only recently in our post World War II world that linguists have begun to consider the social implications of languages, as a result of nation state lines being re-drawn after the war.[3]? Indeed, a recent article from 2017 penned by Mike Medeiros addresses the role of linguistics in conflict, in which he summarizes ?what research has been conducted. He explains that

“[t]he traditional reasons to explain the motives that push ethnic group members to enter into conflict or not, as well as the level of intensity they choose to adopt, have centered on economic inequalities and political differences. However, the magnification of these two major grievances has led collective injustices specifically to ethnicity to be somewhat overlooked. Although ethnic social cleavages are centered on a variety of group markers, sometimes even on multiple ethnic distinctions, the overwhelming majority of ethnic conflicts possess a linguistic difference between the conflicting parties. This suggests that linguistic factors might be a principal element in inter-group tensions. Language-based ethnic tensions are far from uniform. They involve movements that vary greatly in their demands; ranging from the simple desire to protect the group’s language, as with the Frisians in the Netherlands, to the will to form an independent country, as in the case of the Catalans of Spain. They also have a high degree of variance in their intensity with some linguistic groups using peaceful means in order to have their demands acquiesced, such as South Tyroleans in Italy, while others take up arms and resort to violence, a deadly situation exemplified by the Karen in Myanmar. Yet, the role of linguistic factors in intergroup conflicts has been underexplored by scholarly research. One reason for this neglect is due to the fact that studies in political science and sociolinguistics remain disjointed. This situation impedes the understanding of language-based ethnic tensions and has led to a call for more collaboration between these different fields.”[4]

?

In supporting Medeiros’s claims, Marc Shell has explained that “linguistic conflict and cooperation now take place in such a bewildering variety of specific spaces and cases that the importance of language itself is in danger of being lost. Yet, in an age when language is being rediscovered as an ‘historical determinant,’ many wars that were once considered simply ‘religious’ or ‘nationalist’ turn out, on further reflection, to have been ‘linguistic’ as well.”[5]? According to Shell, because so many languages exist and are complex, scholars and society have overlooked the linguistic root of conflict.[6]? Jeroen Darquennes, a professor of German linguistics, supports Shell’s claims and acknowledges the lack of literature on the subject, explaining that “essential for a good understanding of societal language conflict is that in some cases the social or other divisions that in other cases would lead to language conflict either go unnoticed or are not experienced as being problematic.”[7] Dr. Peter Nelde, a leading sociolinguist, published an article in 1987, which addressed language contact and conflict occurring between speakers of languages. He stated that: “conflict plays a role in many social sciences. Linguistically, conflict between different ethnic groups often results from language contact. Problems viewed as political, economic, or sociological in nature are often rooted in linguistic conflict. In the literature, however, contact has overshadowed conflict.”[8]?

Conflict Research Background

??????????? Nils-Christian Bormann, Lars-Erik Cederman, and Manuel Vogt’s study on religion, language, and ethnic war also delves into how conflict researchers will typically adhere to the Barthian school of ethnic thought, which “defines ethnicity in terms of boundaries or cleavages rather than in terms of its specific contents.”[9]? They note how other scholars like political scientists Laitin and Juergensmeyer “have singled out religion as being more conflict-prone than other ethnic dimensions.? In the aftermath of the terrorist attacks of September 11, 2001, this hypothesis has gained considerable currency beyond academia, as illustrated by Tony Blair’s recent assertion that ‘religions difference will fuel this century’s battles.’”[10]? They caution analysts not to put religious conflict or linguistic conflict in separate dimensions. They then conclude that the existing documentation that supports a religious foundation for violence is too biased.[11]? Laitin’s work is notable, however, because he acknowledges language as its own entity.? I disagree with Laitin that “conflict over language is not a prescription for violence.? In fact, under certain potentially incendiary conditions, language conflict can help to contain violence”[12] because of various differences in multiple languages spoken within any given area.

Tristan Mabry shares this opinion and questions Laitin’s conclusions, where he “evaluates quantitative variables ascribed to characteristics of language communities, variables developed by [him] and James Fearon and eventually incorporated into the minorities at risk (MAR) dataset as ancestral language scores (ALS). In sum, such variables are rejected as incapable of representing that which is nominally under consideration: the relative importance of language difference in relation to cultural-cum-national conflict.? Therefore, any conclusions drawn from analyses that employ such data are questionable.”[13] It would be remiss to not give Laitin, a political scientist, credit for trying to understand language conflict; however, he mainly focused on the ancestry of languages to try and quantify differences between them. Laitin brought into question the validity of language data, as respondents are known to not always be truthful when filling out surveys or providing census data.[14] Mabry argues that, “the proponents of language as a critical component of conflict are abundant because the logic of language difference is simple and compelling: ethnic differences lead to ethnic conflict; language differences equal ethnic differences; therefore, language differences are associated with ethnic conflict.”[15] It is a compelling argument, and one that must be acknowledged.

Manuel Vogt’s study of ethnic group mobilization compares overlap between religion and linguistics within ethnic groups, in which he too deviated from Barth’s ambient views of ethnicity.? Vogt looked at “the lasting political impact of distinct ethnic cleavage types related to specific markers, such as race, language, and religion. Specifically, in contrast to the stratifying force of race, linguistic and religious segmentation of ethnic groups increases the risk of violent ethnic conflict. Previous studies that focused on specific ethnic markers also tended to regard religious divisions as particularly violence prone. By contrast, this study’s results suggest that linguistic segmentation might actually be more critical than religious segmentation, especially for separatist violence.”[16]? I sympathize with Vogt’s desire to glean the political impact of these types of conflicts.

Political Background

Anastasia Smirnova and Rumen Iliev shed more light on politicization of language.? They state that,

“[o]ne factor that is particularly relevant for ethnic conflicts but received little empirical attention in studies on language attitudes is the politicization of language, where political views become closely linked to language ideology.? Early proponents of nationalism have essentialized language as the key element of nation building, associating it with purity, authenticity, unity, and historical continuation of a cultural group. […] Since the proliferation of nation-states as a form of governance in the 19th and 20th centuries, in many countries language has become closely related to nationality. In multiethnic societies, official recognition of a minority language often becomes a divisive issue, seen by some as a basic human right, but seen by others as an existential threat to the state. Nationalist politicians, accordingly, recognize the role that language plays in building a group identity and mobilizing supporters, and often make it a central symbol of the political debate.”[17]

?

This too has a trickle effect into the media; if the media are only catering to a select number of language speakers in a given area, this politicization can cause linguistic misrepresentation.[18]? Charles King argues that while many are familiar with the “language of politics,” there has not been much done in regards to ‘”the politics of language,” which is especially notable considering that in certain parts of the world, linguistic status has helped to shape inter-ethnic conflicts or states of harmony.[19]? Indeed, many national governments purposefully discriminate against minority languages, making language control a policy objective.? A perfect example is the continuing debate on whether states will support bilingual learning, as Spanish becomes a more prevalently spoken language within the United States.[20] King warns that “the challenge for security studies specialists and international relations experts, though, is to understand the ways in which language issues can become transformed into life-and-death struggles between contesting ethno-linguistic groups.”[21]

Geographic Background

In addition to sharing a common language, ethnic groups often also share a common homeland or geographic area.[22] Geography has been viewed as both an opportunity for conflict, as well as a motivating factor for conflict.[23] Homelands in particular have “‘the fundamentals of culture and identity. And, as such, [they are] about sustaining cultural boundaries and boundedness…The ‘other’ is always and continuously a threat to the security and integrity of those who share a common home.’ A ‘homeland’ is therefore a special category of territory: it is not an object that can be exchanged, but an indivisible attribute of group identity. This feature explains why ethnic groups rationally view the right to control their homeland as a survival issue, regardless of a territory’s objective value in terms of natural or man-made resources. […] Homeland control ensures that a group’s language can be spoken, its culture expressed, and its faith practiced. This intimate connection between homeland territory and the preservation of identity distinguishes ethnic groups from states.”[24] This phenomenon used to be studied in a non-geographic way through the Minorities at Risk database, but the data was not helpful in a geographic sense, as it was only a compilation of settlement characteristics. In the 1960s, cartographers from the former Soviet Union compiled the Atlas Narodov Mira. Even though the Atlas’s data is outdated, Nils Weidmann still used it, with GIS software, in 2009 to understand ethnic conflicts more clearly.[25]

Weidmann’s research led me to focus on languages spoken in areas of conflict in general. However, unlike other researchers who have focused on drawing comparisons between languages within a certain area, I was more curious as to how many languages are spoken in total within a conflict zone. In a conflict of nation state versus nation state, do the opposing sides have similar linguistic diversity? Does the presence of linguistic diversity have the potential to cause a conflict?

In order to answer these questions, I completed Harvard University’s GIS summer institute and have been working to create maps using conflict data and the languages of the world, with the intention of determining whether or not my research could support or refute previous studies on language conflict. Through mapmaking, I worked to evaluate my contention that language is one of the key elements that needs more consideration when there is conflict in a region. With a greater focus on language, it is possible that conflict resolution would be far easier to achieve. By comparing the conflict area to the number of languages spoken, I hypothesize and will evaluate the following:

·???????? Hypothesis: a country that has multiple languages spoken within its borders is more likely to encounter international or a national conflict than a single language speaking country

In order to evaluate the hypothesis above, I used data from Uppsala University’s Conflict Data Program, provided through their Department of Peace and Conflict Research, established in the 1970s. The Program’s mapping tool displays conflict information from 1975-2020, and the data are? filterable by state-based violence, non-state violence, and one-sided violence. I then joined Uppsala data with linguistic data from Ethnologue, which is recognized as one of the world’s leading linguistic resources.

Although data does exist on conflicts and languages, I wanted to be able to join datasets and project this data comprehensively onto a map.? However, trying to display multiple languages within a particular area of a map is a challenge in GIS.? Trying to “force language into the realm of points, lines, and polygons requires endless compromises between reality and representation.”[26] Weidmann used maps from the 1960s in order to try to map out linguistic differences. It is no wonder monolingual mapping is more the status quo. Indeed, other GIS practitioners from Virginia Polytechnic Institute and State University have stated that: “while linguistic diversity is an integral component of cultural landscapes, the spatial depiction of languages fails to represent all community members. Language is difficult to map and established guidelines are lacking. The perception of power conveyed is arguably the most meaningful design issue of in language mapping, as most language maps inaccurately show one language per place.”[27] Despite these cautions, I was still determined to demonstrate a direct correlation between language and conflict. Therefore, I compromised and? used counts of languages per country, rather than trying to map out what could very well be obsolete observer-reported data.

In joining the data, my hope was to evaluate my contention that language has been a piece of the puzzle that international relations theorists have overlooked, and perhaps discover that there is correlation between linguistic diversity and conflict. I hope that my findings can? open a new pathway of research study for students of international relations.

Chapter III. Definitions

Active Conflict: “A conflict, both state-based and non-state, is deemed to be active if there are at least twenty-five battle-related deaths per calendar year in one of the conflict’s dyads. This rule also applies to settle dyad activity and the activity of the primary warring parties. A secondary warring party is however considered to be active if it actively supports on of the primary parties with regular troops within the state incompatibility. In other words, a secondary warring party does not have to, on its own, incur or suffer twenty-five battle-related deaths to be classified as active. A variant of this coding rule is applied regarding one-sided violence. A one’ sided actor is deemed to be active if an organized group incurs at least twenty-five deliberate killings of civilians in a year.”[28]

?

Actor: “A state or a non-state formally organized or informally organized group.”[29]

?

Armed Conflict: “A state-based armed conflict is a contested incompatibility that concerns government and/or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least 25 battle-related deaths in one calendar year.”[30]

Battle-Related Deaths: “Counted as battle-related deaths is the use of armed force between warring parties in a conflict dyad, be it state-based or non-state, resulting in deaths.”[31]

?

Best Estimate of Deaths (relates to state-based, non-state, one-sided, actors, dyads): “The best estimate consists of the aggregated most reliable number of deaths.”[32]

?

Civilians: “Civilians […] are unarmed people who are not active members of the security forces of the state, or members of an organized armed militia or opposition group. Government officials, such as members of parliament, governors, and councilors, are also excluded and are instead seen as representatives of the government of a state.”[33]

?

Conflict, Interstate (relates to state-based): “A conflict between two or more governments.”[34]

?

Conflict, Intrastate (relates to state-based): “A conflict between a government and a non-governmental party, with no interference from other countries.”[35]

?

Dyad: “A dyad is made up of two armed and opposing actors. In state-based conflicts a dyad is defined as two actors, with one or more being the government, and have a stated incompatibility. In a non-state conflict, a dyad is constructed by at least two organized actors, of which none is the government of a state, which oppose each other with arms. In non-state conflicts it is possible for an alliance of non-state actors to enter a dyad with either an opposing group, or an alliance of opposing groups.”[36]

?

Fatalities: “Deaths incurred in the three categories of organized violence captured by the UCDP. For state-based armed conflict and non-state conflict these are defined as battle-related deaths (i.e., the use of armed force between warring parties in a conflict dyad, be it state-based or non-state, resulting in deaths). For one-sided violence these are deaths stemming from attacks conducted by organized actors, targeting unarmed civilians.[37]

?

Non-State Conflict: “The use of armed force between two organized armed groups, neither of which is the government of a state, which results in at least 25 battle-related deaths in a year.”[38]

?

One-Sided Violence: “The deliberate use of armed force by the government of a state or by a formally organized group against civilians which results in at least 25 deaths in a year.”[39]

?

State: “A state is either an internationally recognized sovereign government controlling a specified territory, or an internationally unrecognized government controlling a specified territory whose sovereignty is not disputed by another internationally recognized sovereign government previously controlling the same territory.”[40]

?

State-Based Armed Conflict: “A state-based armed conflict is a contested incompatibility that concerns government and/or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least 25 battle-related deaths in one calendar year.”[41]

?

Chapter IV. Research Methodology

The most accurate resource containing the language data needed to conduct this research study is Ethnologue’s language and shapefile data.[42]? The Ethnologue dataset that was used for the language shapefiles is from 2004, as that is the most recent version available to obtain from the Harvard Library System. All maps created are based on the 2004 data findings as they were known and captured by Ethnologue at that time. Therefore, this study is based off of the reporting of the number of languages and country boundaries at that time.

The maps I have created were made using ArcGIS?, which is a software created by Esri, the Environmental Systems Research Institute.[43]? My first step was to join the shape file of country boundaries from Ethnologue to the table of all of the world’s languages that existed in 2004. This was called a table join, as I joined a table of the languages to the countries they are spoken in spatially to areas on the Earth’s surface. Throughout my map legends, the number of languages present in any given country are listed and the corresponding colors are presented by country polygon in a choropleth map.? The number of deaths per conflict type are denoted using a graduated point symbology in both the legend and on the maps themselves.[44]?

The next step in my process was a bit complicated, requiring joining the shape country boundaries spatially to the distinct types and number of conflicts that occurred within each year. It is here that my methodology changed. I was originally going to map out several decades worth of conflict and compare interstate to intrastate language conflicts, but due to the difficulty in doing that using ArcGIS?, I determined for the purpose of this study it would be better to map different years before and post 9/11, as the way we view conflict drastically changed after that event.? I chose instead to map the following years: 2004, 2010, 2015, and 2020.

Unlike Weidmann, I was unwilling to use the MAR, GeoEPR, or ANM datasets,[45] as I wasn’t focusing on just minority & ethnic groups, as well as I have mentioned previously, the ANM dataset is outdated. I chose instead to utilize the Uppsala Data Program’s available datasets, which are recognized for their high quality and easy download process for research purposes. I determined I would compare the following datasets in these four particular years:

·???????? UCDP/PRIO Armed Conflict Dataset version 21.1[46][47]

·???????? UCDP Dyadic Dataset version 21.1[48][49]

·???????? UCDP One-Sided Violence Dataset version 21.1[50][51]

·???????? UCDP Non-State Conflict Dataset version 21.1[52][53]

·???????? UCDP Battle-Related Deaths Dataset version 21.1[54]

·???????? Deadly Electoral Conflict Dataset (DECO)[55]

I have previously worked with the Uppsala Conflict Data Program, and I knew that when it came to conflict data, they would have the datasets that I needed to run this study. I also knew that the data would be able to be mapped out, since the program itself has a mapping feature where you can see the conflicts on the world map. Based on my previous experience with this tool, I determined I would not use any other data sources.

The Peace Research Institute Oslo (PRIO) armed conflict dataset was the first dataset I reviewed.? I determined it was appropriate to map that data at the national level, because the government of a state must be one of the acting parties. I was interested to see what data could be gleaned from state vs. a non-state conflicts or conflicts involving two state actors. That this dataset yielded exploratory regression correlation with linguistic diversity validated my initial interest. I want to continue my research with this dataset and map out more prior conflicts to see how the maps change over time.

The Dyadic dataset was the second dataset I reviewed, and I almost discarded it. But, I was curious to see if there would perhaps be different results from the PRIO armed conflict dataset, as they both contained some of the same data from the same data source. This dataset did not provide any correlations, and I will most likely not continue analyzing this information unless something new becomes available.

I next analyzed One-Sided Violence, which involved attacks on civilians. I had previously looked at some of the data before and knew the value of this dataset. The fact that this dataset also yielded correlation with linguistic diversity was of little surprise. I surmised if there were one dataset in particular to have correlation between language and conflict, this would be it. I will be continuing my research with this dataset.

I then reviewed the non-state conflict dataset, because I thought it would be interesting to analyze conflicts where a government was not involved. This particular dataset yielded some correlation on one of the regression tests. This is a dataset I may further explore, because of these preliminary results.

I hoped that the Battle Related Deaths dataset would display a correlation with linguistic diversity since it contains dyad data that was also in the PRIO dataset. However, I was disappointed to find that there were no correlations. I might eventually use this dataset again in the future, but in the interim, I will be focusing on the other datasets that yielded correlations.

Finally, I decided to map out the Deadly Electoral Conflict dataset to see if it would yield any results. This dataset was not of a particular interest to me, but I did not want to discard it without testing it. As this dataset was last completed and published in 2017, I am most interested in it, given the events of the last few years.? I am currently waiting to see the next dataset that will be released and will investigate the correlation with linguistic diversity to see there is any change or not.

I downloaded and saved each dataset, filtered by year, saved the 4 years that I wanted, and spatially joined each of these individual one-year datasets (23 in total) to the shapefile for the countries. This approach enabled me to see how many incidents had occurred during each year within each country. I then joined the language count data to the conflict data, putting both of them together on a map to see if any inferences could be made from visual inspection. Since I had created several tables in the process, I was able to easily join these two new attribute tables together in a table join and see all the data together in one table.

I then wanted to see if I could make any inferences based not only on what could be visually seen on the maps, but from executing exploratory regressions with the new joined datasets. Statistically, we can see that there is some correlation between language diversity and these conflicts, as explained in Chapter VI.

Chapter V. Research Limitations Known Pre-Research Limitations

Despite extensive research, I could find no record of a researcher having previously mapped conflict incidents together with counts of the languages spoken in the conflict zone to evaluate whether or not there is a correlation between the two. While there are datasets containing data regarding conflicts and separate datasets containing language shapefile data for use, combining the data will be a challenge. If my hypotheses are correct, I would like eventually to develop shills with coding and building custom algorithms within ArcGIS and potentially use it as a tool for predicting conflict. I confirm that I will follow all IRB compliance guidelines and submit for approval in the event that I need to use human test subjects.

Post-Research Limitations Acknowledged

During the course of my research, there were the limitations aforementioned concerning a lack of prior research on this topic. Several of these issues have been addressed in previous chapters. Future issues that I can foresee occurring include the possibility that, after mapping out more conflict data, little or no correlations may be found. Another limitation is that there is little to no previous research that has been conducted and recorded in this field, so I do not have access to a different researcher’s work to which I can compare my findings.

The question also remains of what to do with these findings, including ensuring that other students of international relations consider the role of language in conflict and that this approach continues to be utilized to address language conflict.

While in recent years maps have often been criticized as outdated, lacking in creativity, or boring,[56] I am hoping the maps in this research project will not be deemed as such and perhaps someone in the field of linguistic geography might find this work to be of value.

The “field of linguistic geography, in spite of the interdisciplinary overtones of its name, has been practiced primarily by linguists, with limited interaction by geographers. Whereas linguists are interested in the internal workings of language systems, geographers tend to treat language as a unitary variable and have not engaged significantly in research on its internal complexity. This state of affairs was lamented by Wagner (1958) and remained unchanged throughout the remainder of the 20th century in spite of his agitation for more geographer involvement in the sorts of questions that interest historical linguists. Even among linguists, however, linguistic geography is a fractured field. Spatial patterns and geographic relationships are typically secondary to linguists’ other research aims such as diagnosing linguistic relationships, modeling the dynamics of change, or describing structural diversity among the world’s language. As a result, geography is relegated to the periphery of various linguistic subfields more often than it is treated as a unified topic in diachronic linguistics.”[57]

?

Chapter VI. Results

As I was mapping each of the datasets, I saw a pattern emerge: in countries that speak multiple languages, there were conflicts present vs those that only have one language spoken.? I estimated that countries that speak fewer languages are less likely to experience conflict than those that speak more languages. Correspondingly, I presumed that a country with hundreds of languages was far more likely to have conflicts than a country that did not. As you can see from the graduated point symbols within countries, there were zero instances of single language speaking countries being involved in conflicts during these conflict years.

?

UCDP/PRIO Armed Conflict

I was most intrigued by the PRIO armed conflict dataset, which is “a conflict-year dataset with information on armed conflict where at least one party is the government of a state in the time period 1946-2020.”[58][59]? I wanted to see if there would be many instances of conflict involving the government of a state, or if the likelihood of this happening would be lower in a context of more linguistic homogeneity

Figure 1: PRIO Armed Conflict Dataset, 2004

Figure 1 shows that in 2004, India showed the highest number of armed conflicts. The government of India fought against groups of insurgents that year, including the National Democratic Front of Bodoland (NDFB),[60] who were fighting for sovereignty. Of particular note, the NDFB criticized the written script of the Bodo language. The group was primarily composed of Christians, who wanted Roman script to be used, rather than the Devnagri script.[61]? This incident heavily supports the contention that linguistics were a major factor in conflict that year.

?

In comparing Figure 1 and Figure 2, India continued to have armed conflicts, but the number of conflicts decreased in 2010. In evaluating the PRIO dataset, the same actors on both sides of the conflict were still involved.[62][63]

Figure 2: PRIO Armed Conflict 2010

Then in comparing Figure 2 to Figure 3 and evaluating the conflicts of 2010 and 2015, India still had a larger number of armed conflicts, but there was a rise in conflicts in Africa and Southwest Asia. In further exploring the PRIO dataset, we see that these main conflicts occurred in Syria and Mali, both of which saw Islamist armed groups engaged in conflict with the governments of these countries during this time period.[64]?

Figure 3: PRIO Armed Conflict 2015

Between 2015 and 2020, the counts go down in most cases, as Figures 3 and 4 demonstrate, but there is an increase in conflict in the Philippines between the government of the Philippines and the communist party of the Philippines and the Islamic State (IS).[65]? This was at first a bit puzzling at first, but it is important to note that the government of the Philippines has been combating a war on drugs and tensions escalated in 2020, which we can correlate to the COVID-19 pandemic and nationwide lockdown.

Figure 4: PRIO Armed Conflict 2020

As I suspected though, the general pattern in Figures 1-4 is that countries with more languages experienced more conflicts than those that speak fewer languages. This also corresponds to global events. I then wanted to compare this data to the Dyadic Dataset to see if there was a notable difference or not, as both datasets pertain to conflict with one of the actors being a government.

?

UCDP Dyadic Dataset

This dataset is a “dyad-year version of the UCDP/PRIO Armed Conflict Dataset. A dyad consists of two opposing actors in an armed conflict where at least one party is the government of a state.”[66][67]? I expected to see Figures 5-8 look very similar to Figures 1-4, and so I was surprised to see the number of conflicts higher than the PRIO dataset had reported, especially for India.

?

Figure 5: Dyadic Dataset 2004

The conflicts captured in Figure 5 aligned with the same groups as had been active in Figure 1 for India, although there were a few more instances of conflict captured with this dataset.[68]? ??Figure 6 also aligned with Figure 2, again, same year, same groups, remaining consistent.[69]

Figure 6: Dyadic Dataset 2010

An interesting pattern emerges in Figure 7 with four conflicts being prevalent in Mali, the Philippines, and Pakistan. While we know that based on the dataset the same actors were present in Mali and the Philippines as were in the PRIO dataset, it is notable that the conflicts that occurred in Pakistan in 2015 were between the Pakistani Urdu speaking government and different linguistic speaking insurgent groups, and these conflicts were initially begun by tribal members.[70]?

Figure 7: Dyadic Dataset 2015

Sadly, the incidents of violence that occurred between the government of Afghanistan, IS, and the Taliban in 2020 can now be considered foreshadowing for the events of 2021. Figure 8 in this regard is not surprising for Afghanistan. Aligning with the PRIO dataset, the same conflict between the same groups is again recorded for 2020. What is also unsurprising are the number of conflicts recorded below for the Democratic Republic of the Congo (DRC). Again, with the pandemic, it is not surprising that the government of the DRC and rebel groups like the Kata Katanga (which means secede) clashed.[71] The fact that DRC also hosts so many of the world’s internally displaced people, who all speak many different languages, likely is a cause of conflict.

Figure 8: Dyadic Dataset 2020

After reviewing the Dyadic Datasets, and especially the findings of Figure 8, I began to consider what the attacks specifically on civilians look like, in contrast to the actor vs. actor type conflicts that one normally thinks about with conflicts. The human dimension of conflict is often forgotten, and that is something that I hope to research.

?

?

UCDP One-Sided Violence Dataset

This dataset is “an actor-year dataset with information of intentional attacks on civilians by governments and formally organized armed groups.”[72][73]? On a positive note, there were far fewer recordings of violence against civilians than I originally suspected within this dataset. However, I suspect there are actually far more instances of violence against civilians than were recorded or reported. For example, the lack of civilian attacks reported in the Middle East for Figure 9 stood out, especially since this was near the beginning of the Iraq War. A reason for this could very well be a lack of technology during that time, or simply nobody was recording this data. Although this explanation is anecdotal, it is based on what has been seen recently with the Russian invasion and subsequent attacks on Ukrainian civilians in Ukraine in 2022.

??????????? That being said, as seen in the other datasets, it was of no surprise that attacks on civilians in India in 2004 were launched by the NDFB, and other insurgents active within India at that time.[74]

?

Figure 9: One-Sided Violence 2004

A slight change occurred between Figure 9 and Figure 10, highlighting my theory that the lower number of attackswas due to a lack of reporting and lack of technology . India again experienced attacks from insurgents and the NDFB in the fight over Bodoland.[75]

?

Figure 10: One-Sided Violence 2010

In comparing Figure 10 to Figure 11, attacks against civilians in India subsided, while attacks against civilians in the Central African Republic escalated. These attacks were by the Lord’s Resistance Army, anti-Balaka, Popular Front for the Rebirth of the Central African Army, Union for Peace in the Central African Army, and Democratic Front for the People of Central Africa,[76] suggesting these attacks could be for religious or political reasons. However, it is of note that there are over 54-105 different languages spoken in this particular country, suggesting that the large number of languages could have contributed to these escalations.

Figure 11: One-Sided Violence 2015

Figure 12 shows a shift from the Central African Republic to Ethiopia, where the governments of Ethiopia and ethnic militias from Eritrea launched attacks on civilians. Ethiopia’s people also speak many different languages, and with ethnic troops attacking from a neighboring country, I can state with near certainty that linguistic divide has contributed to these conflicts.

Figure 12: One-Sided Violence 2020

With this dataset in particular, there were zero instances of conflicts in countries, including one-sided violence, with fewer than ten languages.

?

UCDP Non-State Conflict

?

The Non-State Conflict dataset is “a conflict-year dataset containing information on communal and organized armed conflict where none of the parties is the government of a state.”[77][78]? For this dataset, my initial hypothesis was to anticipate seeing larger numbers of terrorist organization activity. While incorrect, the number of incidents are higher than the other previously researched datasets, showing there has been a higher incidence of attacks from non-governmentally affiliated groups.

??????????? Figure 13 was surprising in the fact that there was no conflict data for India, as found in the previous datasets. The violence in Somalia at the time is attributed to clans attacking other clans, which again, given the high number of different languages spoken in Somalia, supports my linguistic theory. The incidents in Mexico were due to different cartels attacking other cartels.[79]?

Figure 13: Non-State Conflict 2004

In Figure 14, the greatest number of conflicts are in the same countries again in 2010. In Mexico, it is still inter-cartel conflict. For Somalia, there is more conflict between the State of Somalia, the mujahideen, and the Islamic party, rather than more of the clan conflict that was recorded in 2004.[80]

Figure 14: Non-State Conflict 2010

Figure 15 shows the highest number of conflicts thus far in any dataset. The conflicts in Mexico for this year were the same as in 2004 and 2010, being attributed to the cartel wars. The conflicts in Nigeria were between different ethnic groups and opposing Christians and Muslims. Syria’s multiple conflicts are a direct result of the civil war and fighting with IS.[81]

Figure 15: Non-State Conflict 2015

Based on the trends for Figures 13-15, Figure 16 was not surprising, as the conflicts between the cartels in Mexico continued, as well as for the ethnic groups of Nigeria.[82]

Figure 16: Non-State Conflict 2020

Based on this current trajectory, I can predict with some confidence that we will see more instances of conflict within Mexico and Nigeria within the next 5 years, and I expect the release of the 2025 data to confirm this prediction.

?

UCDP Battle Related Deaths

I was most curious about this dataset, out of any of the datasets reviewed. The “file contains a dyad-year dataset with information on the number of battle-related deaths in the conflicts from 1989-2020 that appear in the UCDP/PRIO Armed Conflict Dataset.”[83]? I was curious to see where the battles took place and what languages were spoken in the area of conflict.

Figure 17: Battle-Related Deaths 2004

Knowing much of this data also corresponds to the PRIO dataset, Figure 17 did not come as a surprise. India was unsurprising, with battles between the government of India and insurgents in Bodoland, Manipur, Kashmir, Assam, and Tripura.[84]

Figure 18: Battle-Related Deaths 2010

Figure 18 showed relatively minor changes that reflect the PRIO data, although ?the number of conflicts between the government of India and the insurgent groups declined during this period of time.[85]

Figure 19: Battle-Related Deaths 2015??????????????????????????????

The year 2015? had a higher number of recorded conflict incidents that occurred globally as seen in Figure 19, which also correlated with the PRIO dataset for the countries of Mali, the Philippines, Nigeria, and India.[86]? Figure 20 below aligned as well, with no surprising new conflict areas that differed from the PRIO dataset.[87]

Figure 20: Battle-Related Deaths 2020

If anything, it will be interesting to see what the 2025 data brings in comparison to the 2020 recorded data as seen above. It is notable that there are few battle-related deaths in North or South America in the years mapped.? It may be worthwhile to obtain and review data prior to 2004 for these regions.

?

Deadly Electoral Conflict

In terms of conflict, I was unsure what results this dataset in particular would yield. It is “a global, georeferenced event dataset, based on UCDP data, identifying electoral violence with lethal outcomes from 1989 to 2017.”[88]? As data is only available up to the year 2017, there is missing data for 2020, resulting in only three figures for this dataset.

Figure 21 reflected an alarming rate of electoral violence. In India, the violence was often between the Government of India and Kashmir insurgents, as well as Kashmir insurgents and civilians as a retaliatory measure. In Afghanistan, there is an analogous situation with the Government of Afghanistan and the Taliban, with resulting Taliban violence toward civilians. In both instances, the governments of these countries were not aggressors against civilians.[89]

Figure 21: Deadly Electoral Conflict 2004

In Figure 22, there is an increase in Deadly Electoral Conflict particularly in Afghanistan and the Philippines. This dataset includes headlines for these incidents, and they are all repetitive. In Afghanistan, there were a number of blasts, roadside bombs, rocket attacks, and gunmen who killed civilians and election candidates. For the Philippines, most of the violence was between warring clans, and involving opposition to the election of different candidates in certain villages.[90]?

Figure 22: Deadly Electoral Conflict 2010

In Figure 23 there is a distinct distinct set of Electoral Conflicts that emerge for 2010, which are vastly different than those seen in Figures 21 and 22.? The number of conflicts seen in Bangladesh was due to political violence. Burundi’s conflicts were between the Government of Burundi and opponents of Nkurunziza, as well as targeted killings of civilians in a genocidal sweep. In all instances, there were electoral candidates and civilians both targeted.[91]?

Figure 23: Deadly Electoral Conflict 2015

I am interested in reviewing the reports of violence for 2020 when the data is available in order to see if there is any further trending data, or if electoral conflict receded in 2020. Given the data gleaned from the other datasets, however, I predict that we will see prominent levels of electoral conflict here as well.

?

Regressions

After mapping the above datasets, I wanted to know if I could establish through exploratory regressions whether there was any significance in the data that I had plotted on the map, other than the inferences drawn from viewing the maps, and looking at the raw data and making conclusions based on global events.

I combined all of the sums for each conflict measures over all four years in the study . languages, and the country shape data with the conflicts and languages into one massive join, and then projected this information into a Cylindrical Equal Area Projection.

In running the exploratory regression tool, I set the number of languages as the dependent variable and the sum of the variables as candidates. The results are as follows:

??????????? In looking at this, one-sided violence, PRIO armed conflict, and non-state conflicts showed single-variable correlation. What was most interesting to me was the K(BP).

The K(BP) “Koenker (BP) Statistic (Koenker's studentized Bruesch-Pagan statistic) is a test to determine whether the explanatory variables in the model have a consistent relationship to the dependent variable both in geographic space and in data space. When the model is consistent in geographic space, the spatial processes represented by the explanatory variables behave the same everywhere in the study area (the processes are stationary). When the model is consistent in data space, the variation in the relationship between predicted values and each explanatory variable does not change with changes in explanatory variable magnitudes (there is no heteroscedasticity in the model). Suppose you want to predict crime, and one of your explanatory variables is income. The model would have problematic heteroscedasticity if the predictions were more accurate for locations with small median incomes than they were for locations with large median incomes. The null hypothesis for this test is that the model is stationary. For a 95 percent confidence level, a p-value (probability) smaller than 0.05 indicates statistically significant heteroscedasticity and/or nonstationarity. When results from this test are statistically significant, consult the robust coefficient standard errors and probabilities to assess the effectiveness of each explanatory variable.”[92]?

?

I then reversed the dependent and independent variables to see if the correlations would change. In setting the number of languages to the independent variable and the sum of the variables to be univariate, there were the following notable correlations for both one-sided violence and the PRIO armed conflict dataset, but not for non-state conflicts. I suspect that non-state conflict did not show notable correlations because there is not governmental involvement, but more research into this matter is necessary.

?

One-Side Violence results:

PRIO Armed Conflict Results:

?

Non-State Conflict Results:

Chapter VII. Conclusion

As evidenced in the maps above, the corresponding information from the datasets, and the exploratory regressions, I have established a correlation between language diversity and conflict, which supports my hypothesis in that the number of languages that are spoken between opposing sides in a conflict can be a cause of the conflict. There were no instances of conflict in countries whose people speak ten languages or fewer within the chosen timeframe, other than Haiti, which is notable. That absence also supports my hypothesis that countries that speak more than one language are more likely to experience conflict.

In some instances, seeing the conflicts on the maps can potentially help predict future conflicts, as was seen with Afghanistan. In continuing my work, my first step will be to run a comparative analysis of all data between 2000-2020. Because I know there were changes in the several year spans that I had bypassed in this analysis, I am especially interested to see the yearly changes on the map, and to join all of the yearly data together to see if my exploratory regression findings are altered. I am also keen in mapping other conflicts from previous decades as well and seeing if my results may perhaps shed new light on past historic events.

I also would like to take a more micro approach and map where different insurgent groups and terrorist organizations originate from within countries, to compare with the location of the civilians and governments that they attack. Such mapping could determine if a common language is spoken within that particular region of the country. This would further support my findings.

My initial findings further support the importance of foreign language education and linguistic learning in all levels of education. Ultimately, these findings show that there is another element – linguistic diversity – that is important to consider when studying conflict, and that languages can be at the root of conflicts.

?

References

Anderson, Christopher & Aida Paskeviciute (2006)? How Ethnic and Linguistic Heterogeneity Influence the Prospects for Civil Society: A Comparative Study of Citizenship Behavior. The Journal of Politics, Vol. 68, No. 4, November 2006, 787.

Bormann, Nils-Christian, Lars-Erik Cederman, and Manuel Vogt (2017) Language, Religion, and Ethnic Civil War.? Journal of Conflict Resolution,? 744-745.

Darquennes, Jeroen. “Language conflict research: A state of the art.” International Journal of the Sociology of Language, (2015) 235: 7-32.

Duffy Toft, Monica (2002) Indivisible territory, geographic concentration, and ethnic war. Security Studies, 12:2, 87.

Eck, Kristine & Lisa Hultman (2007) Violence Against Civilians in War. Journal of Peace Research 44(2).

Esri. All rights reserved. www.esri.com.

Esri. “How OLS regression works,” web, https://pro.arcgis.com/en/pro-app/2.8/tool-reference/spatial-statistics/how-ols-regression-works.htm.

Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and H?vard Strand (2002) Armed Conflict 1946-2001: A New Dataset. Journal of Peace Research 39(5).

Fjelde, Hanne and Kristine H?glund (2021) “Introducing the Deadly Electoral Conflict Dataset (DECO)” Journal of Conflict Resolution,?https://doi.org/10.1177/00220027211021620.

Glottolog. “Languages.” Languages. 2021. https://glottolog.org/glottolog/language.

Harbom, Lotta, Erik Melander & Peter Wallensteen (2008) Dyadic Dimensions of Armed Conflict, 1946-2007. Journal of Peace Research 45(5).

Haynie, Hannah J. (2014) Geography and Spatial Analysis in Historical Linguistics. Language and Linguistics Compass, 344.

Iberhard, David M., Gary F. Simons, and Charles D. Fenning (eds). Ethnologue: Language of the World. Dallas, Texas: SIL International. 2004.

?

King, Charles King (1997) Policing Language: Linguistic Security and the Sources of Ethnic Conflict. Security Dialogue, Vol. 28(4), 494.

?

Kisembe, Denis. “Language, culture and conflict resolution.” Research Paper, 2018. https://www.grin.com/document/433194

Laitin, David D. Laitin? (2000) Language conflict and violence: the straw that strengthens the camel’s back. Arch.europ.sociol., XLI,? 98.

?

Lake, Nell. “Language Wars.” Harvard Magazine, March-April 2002.

Mabry, Tristan James (2010) Language and Conflict. International Political Science Review, 190.

?

Luebbering, Candice R., Korine N. Kolivras & Stephen P. Prisley Visualizing Linguistic Diversity Through Cartography and GIS. The Professional Geographer, 65:4, 581.

?

Nelde, Peter H. “Language contact means language conflict.” Journal of Multilingual and Multicultural Development, (1987) 8:1-2, 33-42.

Newmeyer, Frederick J. “The History of Modern Linguistics,” Linguistic Society of America, 2022, https://www.linguisticsociety.org/resource/history-modern-linguistics

Medeiros, Mike. “The Language of Conflict” Ethnicities, (October 2017), Vol.17, No.5, pp. 627-645.

?

Mpofu, Philip & Davie E. Mutasa (2014) Language policy, linguistic hegemony and exclusion in the Zimbabwean print and broadcasting media. South African Journal of African Languages, 34:2, 226.

?

Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

Shell, Marc. “Language Wars.” CR: The New Centennial Review 1, no.2 (2001): 1-17.

?

Smirnova, Anastasia and Rumen Iliev (2017) Political and Linguistic Identities in an Ethnic Conflict. Journal of Language and Social Psychology, Vol. 36(2) 211-225.

?

South Asia Terrorism Portal. “National Democratic Front of Bodoland.” South Asia Terrorism Portal, 2001. https://www.satp.org/satporgtp/countries/india/states/assam/terrorist_outfits/ndfb.htm.

Sundberg, Ralph, Kristine Eck, and Joakim Kreutz (2012) Introducing the UCDP Non-State Conflict Dataset. Journal of Peace Research 49(2).

Uppsala Conflict Data Program. “UCDP.” Department of Peace and Conflict Research. Accessed April 30, 2021. https://ucdp.uu.se/.

Uppsala University. “UCDP Definitions.” Definitions. Accessed April 30, 2021. https://www.pcr.uu.se/research/ucdp/definitions/#tocjump_8829967763543327_2

Vogt, Manuel (2018) Ethnic Stratification and the Equilibrium of Inequality: Ethnic Conflict in Postcolonial States. International Organization 72, 133.

Weidmann, Nils B (2009) Geography as Motivation and Opportunity.? Journal of Conflict Resolution, Volume 53 Number 4, 526-541.

?

[1] Denis Kisembe, “Language, culture and conflict resolution.” Research Paper, 2018. https://www.grin.com/document/433194.

[2]? Christopher J. Anderson & Aida Paskeviciute, “How Ethnic and Linguistic Heterogeneity Influence the Prospects for Civil Society: A Comparative Study of Citizenship Behavior,” The Journal of Politics, Vol. 68, No. 4, November 2006, 787.

[3] Frederick J. Newmeyer, “The History of Modern Linguistics,” Linguistic Society of America, 2022, https://www.linguisticsociety.org/resource/history-modern-linguistics.

[4] Mike Medeiros, “The Language of Conflict” Ethnicities, (October 2017), Vol.17, No.5, pp. 627-645.

[5] Marc Shell, “Language Wars,” CR: The New Centennial Review 1, no.2 (2001):1-17.

?

[6] Nell Lake, “Language Wars,” Harvard Magazine, March-April 2002.

?

[7] Jeroen Darquennes, “Language conflict research: A state of the art,” International Journal of the Sociology of Language, (2015) 235: 7-32.

?

[8] Peter H. Nelde, “Language contact means language conflict,” Journal of Multilingual and Multicultural Development, (1987) 8:1-2, 33-42.

[9] Nils-Christian Bormann, Lars-Erik Cederman, and Manuel Vogt, “Language, Religion, and Ethnic Civil War,” Journal of Conflict Resolution, 2017, 744-745.

[10] Ibid., 745.

[11] Ibid., 764-766.

[12] David D. Laitin, “Language conflict and violence: the straw that strengthens the camel’s back,” Arch.europ.sociol., XLI, 2000, 98.

[13] Tristan James Mabry, “Language and Conflict,” International Political Science Review, 2010,? 190.

[14] Ibid., 195.

[15] Ibid., 191.

[16] Manuel Vogt, “Ethnic Stratification and the Equilibrium of Inequality: Ethnic Conflict in Postcolonial States,” International Organization 72, Winter 2018, 133.

[17] Anastasia Smirnova and Rumen Iliev, “Political and Linguistic Identities in an Ethnic Conflict,” Journal of Language and Social Psychology, Vol. 36(2) 211-225, 2017, 213.

[18] Philip Mpofu & Davie E. Mutasa, “Language policy, linguistic hegemony and exclusion in the Zimbabwean print and broadcasting media,” South African Journal of African Languages, 34:2, 2014, 226.

[19] Charles King, “Policing Language: Linguistic Security and the Sources of Ethnic Conflict,” Security Dialogue, Vol. 28(4), 1997, 494.

[20] Ibid., 496.

[21] Ibid.

[22] The author of thesis acknowledges that people are often displaced and that diasporas exist.

[23] Nils B Weidmann, “Geography as Motivation and Opportunity,” Journal of Conflict Resolution, Volume 53 Number 4, August 2009, 529.

[24] Monica Duffy Toft, “Indivisible territory, geographic concentration, and ethnic war,” Security Studies, 12:2, 2002, 87.

[25] Nils B Weidmann, “Geography as Motivation and Opportunity,” Journal of Conflict Resolution, Volume 53 Number 4, August 2009, 526-541.

[26] Candice R. Luebbering, Korine N. Kolivras & Stephen P. Prisley, “Visualizing Linguistic Diversity Through Cartography and GIS,” The Professional Geographer, 65:4, 581.

[27] Ibid., 580.

[28] “UCDP Definitions,” Definitions, Uppsala University, accessed April 30, 2021, https://www.pcr.uu.se/research/ucdp/definitions/#tocjump_8829967763543327_2.

?

[29] Ibid.

?

[30]Ibid.

?

[31] Ibid.

?

[32] Ibid.

?

[33] Ibid.

?

[34] Ibid.

?

[35] Ibid.

?

[36] Ibid.

?

[37] Ibid.

?

[38] Ibid.

?

[39] Ibid.

?

[40] Ibid.

?

[41] Ibid.

[42] Iberhard, David M., Gary F. Simons, and Charles D. Fenning (eds). 2004.? Ethnologue: Language of the World. Dallas, Texas: SIL International.

[43] Esri. 2022. All rights reserved. www.esri.com.

?

[44] Iberhard, David M., Gary F. Simons, and Charles D. Fenning (eds). 2004.? Ethnologue: Language of the World. Dallas, Texas: SIL International.

[45] https://journals-sagepub-com.ezp-prod1.hul.harvard.edu/doi/pdf/10.1177/0738894210393217

[46] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[47] Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and H?vard Strand (2002) Armed Conflict 1946-2001: A New Dataset. Journal of Peace Research 39(5).

?

[48] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[49] Harbom, Lotta, Erik Melander & Peter Wallensteen (2008) Dyadic Dimensions of Armed Conflict, 1946-2007. Journal of Peace Research 45(5).

?

[50] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[51] Eck, Kristine & Lisa Hultman (2007) Violence Against Civilians in War. Journal of Peace Research 44(2).

?

[52] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[53] Sundberg, Ralph, Kristine Eck, and Joakim Kreutz (2012) Introducing the UCDP Non-State Conflict Dataset. Journal of Peace Research 49(2).

[54] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[55] Fjelde, Hanne and Kristine H?glund (2021) “Introducing the Deadly Electoral Conflict Dataset (DECO)” Journal of Conflict Resolution,?https://doi.org/10.1177/00220027211021620

[56] Candice R. Luebbering, Korine N. Kolivras & Stephen P. Prisley, “Visualizing Linguistic Diversity Through Cartography and GIS,” The Professional Geographer, 65:4, 581.

[57] Hannah J. Haynie, “Geography and Spatial Analysis in Historical Linguistics,” Language and Linguistics Compass, 2014, 344.

[58] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[59] Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and H?vard Strand (2002) Armed Conflict 1946-2001: A New Dataset. Journal of Peace Research 39(5).

[60] Ibid.

?

[61] “National Democratic Front of Bodoland,” South Asia Terrorism Portal, 2001.? https://www.satp.org/satporgtp/countries/india/states/assam/terrorist_outfits/ndfb.htm.

?

[62] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[63] [63] Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and H?vard Strand (2002) Armed Conflict 1946-2001: A New Dataset. Journal of Peace Research 39(5).

[64] Ibid.

[65] Ibid.

[66] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[67] Harbom, Lotta, Erik Melander & Peter Wallensteen (2008) Dyadic Dimensions of Armed Conflict, 1946-2007. Journal of Peace Research 45(5).

[68] Ibid.

[69] Ibid.

[70] Ibid.

[71] Ibid.

[72] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[73] Eck, Kristine & Lisa Hultman (2007) Violence Against Civilians in War. Journal of Peace Research 44(2).

[74] Ibid.

[75] Ibid.

[76] Ibid.

[77] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

?

[78] Sundberg, Ralph, Kristine Eck, and Joakim Kreutz (2012) Introducing the UCDP Non-State Conflict Dataset. Journal of Peace Research 49(2).

[79] Ibid.

[80] Ibid.

[81] Ibid.

[82] Ibid.

[83] Pettersson, Therese, Shawn Davis, Amber Deniz, Garoun Engstr?m, Nanar Hawach, Stina H?gbladh, Margareta Sollenberg & Magnus ?berg (2021). Organized violence 1989-2020, with a special emphasis on Syria. Journal of Peace Research 58(4).

[84] Ibid.

[85] Ibid.

[86] Ibid.

[87] Ibid.

[88] Fjelde, Hanne and Kristine H?glund (2021) “Introducing the Deadly Electoral Conflict Dataset (DECO)” Journal of Conflict Resolution,?https://doi.org/10.1177/00220027211021620.

[89] Ibid.

[90] Ibid.

[91] Ibid.

[92] Esri, “How OLS regression works,” web, https://pro.arcgis.com/en/pro-app/2.8/tool-reference/spatial-statistics/how-ols-regression-works.htm.

?