INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

Arabic Language Attrition and State Stability: Empirical Validation

of the SPH-LENS Framework in the Arab World

Mostafa Ahmed., Dr. Mohamed Shadi

Al Habtoor Research Centre – Dubai

DOI: https://dx.doi.org/10.47772/IJRISS.2025.91100355

Received: 28 November 2025; Accepted: 05 December 2025; Published: 11 December 2025

ABSTRACT

Arabic’s gradual attrition has been theorised as a potential national security threat, but this claim has not

previously been tested empirically. Building on our earlier conceptual work – the Socio-Political-Historical

(SPH) framework and the SPH-LENS early warning system – the paper examines whether erosion in Arabic

language vitality is associated with rising state fragility across the 22 Arab League states. Using a panel dataset

covering roughly 2000–2025, the paper constructs a composite Arabic Attrition Index (AAI), operationalising

SPH LENS indicators, and compare it to national stability measures such as the Fragile States Index (FSI). Fixed

effects panel regressions, panel Granger causality tests, and robustness checks with economic, demographic, and

institutional controls are employed to isolate the language factor. The paper finds that declines in Arabic’s

societal role – particularly in education, science, and media – significantly predict subsequent increases in state

fragility, even after accounting for confounders. These results provide the first quantitative evidence that

language attrition and instability are linked, reframing Arabic language policy as a strategic rather than purely

cultural concern. The paper concludes with policy recommendations for Arab governments and the Arab League

and outlines avenues for further research on language vitality as an early warning indicator of national cohesion

and security.

Keywords: Arabic, Arab League, Attrition Index, language Security, State Fragility, Language–security nexus,

National Cohesion, Sociolinguistic Stratification

INTRODUCTION

Arabic is officially celebrated as one of the world’s most robust languages, yet recent scholarship highlights its

declining societal role and the potential security implications of this shift. Our previous work argued that the

erosion of Arabic – evident in shrinking functional domains and prestige – constitutes a first-order threat to Arab

national security, with possible outcomes including fragmentation reminiscent of the former Yugoslavia In this

view, language decline is not merely a cultural loss: when a shared lingua franca weakens, the risk increases that

societies fracture along ethnic, sectarian or regional lines, undermining national unity and stability.

Arabs without Arabic introduced the Socio-Political-Historical (SPH) framework, drawing on Bourdieu’s

concept of linguistic capital and Gramsci’s notion of cultural hegemony, to explain how global linguistic power

dynamics erode Arabic’s status. That work projected a long-term decline in Arabic vitality across 22 Arab

countries, suggesting that some could fall below a critical language viability threshold within decades. SPH

LENS (Socio-Political-Historical Language Early warning & National security System) extended this

framework by organising measurable indicators into three dimensions – Socioeconomic, Political and Historical

– to generate composite risk scores for Arabic attrition. The paper showed conceptually how such an index could

“red flag” trends like a shift to English or French as media of instruction as early signs of broader sociolinguistic

displacement and potential social fissures.

Despite these advances, a crucial gap remained: no empirical validation had yet demonstrated that Arabic

language attrition correlates with, let alone precedes, national instability. Prior discussions of Arabic’s decline

as a security threat rested largely on case studies, historical analogy and theoretical reasoning. This left a core

question unanswered: does the loss of Arabic’s societal functions and prestige measurably increase the risk of

Page 4529

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

state fragility and internal conflict in the Arab world? Answering this question is both an academic imperative –

to ground theory in data – and a policy imperative, since a positive finding would justify incorporating language

vitality into national security monitoring alongside economic, social and military indicators.

This paper addresses that gap. we extend the SPH/SPH LENS frameworks into a testable hypothesis and bring

quantitative evidence to bear on the language–security nexus. We briefly review relevant literature on language

vitality, identity and stability; describe our research design, including the construction of an Arabic Attrition

Index (AAI) and the selection of fragility indicators; and present results from panel fixed effects regressions and

Granger causality tests. We then interpret the findings in light of sociolinguistic and political theory, discuss

their implications for Arab governments and the Arab League, and outline a proactive “language security”

agenda. We conclude by summarising our contributions and highlighting directions for future research.

LITERATURE REVIEW

This review situates the SPH LENS framework within three intersecting literatures: (i) language vitality and

shift as socio-political barometers; (ii) the relationship between language, identity and state cohesion; and (iii)

efforts to operationalise dynamic, early warning indicators of linguistic change. Rather than disputing existing

endangerment classifications, we argue for complementing them with tools that capture gradual reallocations of

linguistic capital across high-stakes domains—changes that may have consequences for national cohesion well

before a language is formally “endangered”.

Language Vitality as a Socio-Political Barometer

Global assessments such as UNESCO’s Atlas of the World’s Languages in Danger and the Ethnologue/EGIDS

scale classify Arabic as “safe”, given its large speaker base, formal status in 22 states and centrality to religious

practice.ⁱThese frameworks, drawing on Fishman’s pioneering work on intergenerational transmission and

domain loss, are invaluable for identifying threatened minority languagesⁱⁱ. However, they are less sensitive to

more subtle shifts in where and how a nominally secure language is used—especially in elite, high-value domains

such as science, higher education, business and digital media.

Sociolinguistic and political-economy approaches emphasise that language vitality is embedded in broader

structures of power and incentive. Bourdieu’s notion of linguistic capital treats language varieties as unequally

valued resources, whose “market” value is determined by pay-offs in education, labour markets and social

mobilityⁱⁱⁱ. Gramsci’s concept of cultural hegemony similarly highlights how dominant languages help naturalise

world-views, aspirations and hierarchies^iv. Extend these insights to global language hierarchies, showing how

English has become tied to globalisation, scientific prestige and access to transnational networks.^v

Within the Arab world, a growing body of work documents precisely this kind of stratified multilingualism.

Suleiman shows how Arabic functions simultaneously as communicative medium and ideological symbol of

Arab nationalism and belonging, while also noting how its role is contested in specific national settings.^viThe

Arab Thought Foundation’s Arabic Language Report similarly presents Arabic as a core component of collective

identity but warns of erosion in education, culture and media under globalising pressures. More recent studies

chart the expanding role of English (and in some contexts French) as a language of higher education, business

and technology in the Gulf and beyond, often at the expense of Arabic in advanced knowledge production and

academic publishing.^vii

These studies converge on a common pattern: Arabic is rarely displaced outright, but its functional profile

changes. It is retained, and often celebrated, in symbolic, religious and low-stakes communicative arenas, while

foreign languages gain ground in domains that confer status and opportunity. Research on “Arabizi” and hybrid

Arabic–Latin scripts point in the same direction, with anxieties expressed that such practices may weaken links

to Classical and Modern Standard Arabic and, by extension, to Arab identity.^viii

A constructive reading of this literature suggests that the key issue is not whether Arabic is “endangered” in a

conventional sense, but whether incremental reallocations away from Arabic in elite and strategic domains

constitute a socio-political barometer—signalling deepening social stratification, widening informational divides

Page 4530

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

and shifting prestige structures. What is largely missing is a systematic, cross-national attempt to quantify these

reallocations over time and to test whether they carry observable consequences for state fragility.^ix

Language, Identity and State Cohesion

A second body of work links language to nation-building, identity and the cohesion of modern states. Classic

accounts of nationalism by Anderson and Gellner view shared, standardised languages as central to the

emergence of “imagined communities”: they enable mass communication, schooling and bureaucratic

integration, allowing geographically dispersed populations to perceive themselves as part of the same political

community.^{x xi}From this vantage, language is not just a marker of identity but part of the infrastructure through

which states cultivate solidarity and legitimacy.

In the Middle East, Suleiman shows how Arabic has been mobilised as both a unifying symbol of pan-Arab

nationalism and a site of intra-Arab ideological contestation.^xiiGreenberg’s analysis of the former Yugoslavia

illustrates how the deliberate codification of closely related varieties into distinct “languages” can crystallise

political cleavages and contribute to state disintegration.^xiiiComparative work on language policy and ethnic

conflict similarly argues that decisions over official languages, medium of instruction and language rights can

either mitigate or exacerbate tensions in multilingual polities. In extreme cases, language policies have been

shown to operate as tools of domination or exclusion, fuelling grievances and, at times, violent mobilization.^xiv

The quantitative civil-war literature reinforces the idea that linguistic and ethnic cleavages can shape the risk of

insurgency, especially when combined with weak state capacity. Fearon and Laitin, for example, highlight how

rough terrain, low income and state weakness create opportunities for insurgent groups mobilised along ethnic

and linguistic lines.^xvWhile they caution against simple mechanical links between fractionalisation and conflict,

subsequent work has shown that politicised linguistic boundaries can heighten the risk of instability, particularly

where linguistic minorities are excluded from state institutions or denied language rights.^xvi

Yet this literature focuses overwhelmingly on minority languages and multilingual states in which a dominant

national language is seen as a tool for integration—or, conversely, as a vehicle of assimilation. Far less attention

has been paid to cases where a historically central national or religious lingua franca—such as Arabic—appears

to be losing ground in high-value domains to external global languages. Recent Arab scholarship and policy

reports warn that such a process could widen the gap between globally connected, foreign-language elites and

largely Arabic-speaking publics, fragment public spheres and weaken shared frames of reference.^xviiBut these

arguments remain largely qualitative. No large-N, cross-national studies have tested whether measurable erosion

in Arabic’s societal role is associated with changes in widely used indicators of state fragility. Addressing this

gap is a core contribution of the present article.

From SPH to SPH-LENS: Operationalising a Dynamic Early-Warning System

A third strand of scholarship, closer to sociolinguistics and language policy, develops tools for assessing

language vitality and planning interventions. UNESCO’s Language Vitality and Endangerment guidelines

propose a multi-factor framework—covering intergenerational transmission, domains of use, response to new

media, institutional support and community attitudes—to inform documentation and policy priorities.^xviii

Fishman’s (1991) Graded Intergenerational Disruption Scale similarly offers a staged model of language shift

and recovery. These tools, and their subsequent refinements, have proved highly influential in evaluating the

status of local and minority languages and in designing maintenance or revitalisation programmes.^xix

However, existing frameworks have three limitations from a national-security perspective. First, they are

typically applied in cross-section or at long intervals, rather than as annual, country-level time series that can

feed into early-warning systems. Second, they focus primarily on risk of language death—that is, on whether a

speech community will maintain intergenerational transmission—not on more subtle but politically salient

reallocations of language functions within states where the language remains numerically dominant. Third, they

are rarely linked empirically to macro-political outcomes such as state fragility, civil conflict or regime stability.

Page 4531

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

The SPH framework, introduced in our earlier work, addresses these gaps by treating language vitality explicitly

as a socio-political variable embedded in Socioeconomic, Political and Historical structures.^xxSPH-LENS

(Socio-Political-Historical Language Early-warning & National-security System) extends this conceptual model

into an operational architecture. It organises observable indicators into three dimensions—socioeconomic (S),

political-institutional (P) and historical-structural (H)—with the explicit aim of generating composite “risk

scores” that can be tracked yearly across countries. Socioeconomic indicators capture incentives and usage in

education, knowledge production and media (e.g. language of instruction in secondary and tertiary education,

share of scientific output and patents in Arabic, Arabic content in broadcast and digital platforms). Political

indicators capture formal status and institutional backing (e.g. constitutional provisions, language-law reforms,

language-planning bodies and budgetary support). Historical indicators proxy deeper legacies, such as colonial

language regimes and the size and visibility of non-Arabic linguistic communities.

In this article we instantiate SPH-LENS in the form of an Arabic Attrition Index (AAI), a composite measure

designed for country-year tracking and cross-national comparison. Higher AAI values denote greater erosion of

Arabic’s societal role, as reflected in domain shifts toward foreign languages or colloquial varieties and in weaker

institutional support for Modern Standard Arabic. Because the indicators that feed into the AAI sit “upstream”

of intergenerational break-down, movements in the index—such as reductions in Arabic-medium university

provision or declines in Arabic digital content—are conceived as leading indicators that may surface several

years before conventional endangerment metrics would register a problem.

This positioning aligns SPH-LENS with a broader literature on structural early-warning systems in conflict and

fragility studies, which relies on composite indices such as the Fragile States Index, the Worldwide Governance

Indicators and related tools to monitor risk.^xxiWhat is novel here is the integration of a language-based

early-warning index into this architecture. To our knowledge, no prior study has (i) constructed a panelised,

cross-national index of Arabic language attrition grounded in sociolinguistic theory and (ii) systematically tested

its association with standard measures of state fragility using longitudinal econometric techniques. The empirical

sections that follow take up this task.

METHODOLOGY

To investigate the link between Arabic language attrition and national stability, we design a comparative

longitudinal study covering the 22 member states of the Arab League. The analysis employs a panel dataset in

country-year format, allowing us to exploit both cross-country and over-time variation. We focus on the period

approximately 2000–2025 (subject to data availability), a span that captures the post-globalization acceleration

of English/French penetration in the Arab world as well as significant political developments (e.g., the Arab

Spring and its aftermath). This timeframe provides enough temporal variation to conduct tests of causal ordering,

while the inclusion of all Arab states offers a broad comparative perspective.

Dependent variable: State Fragility and Instability

Our primary dependent variable is a measure of state stability (or lack thereof). We operationalise this using the

Fragile States Index (FSI) published annually by the Fund for Peace. The FSI is a widely used composite

indicator that assesses a country’s vulnerability to collapse or conflict, aggregating 12 political, social, and

economic components (grouped into Cohesion, Economic, Political, and social categories) into an overall

fragility score. Higher scores on the FSI indicate greater fragility and risk of instability, whereas lower scores

indicate more stability. We obtain annual FSI scores for each Arab country throughout the study period. In

addition to the overall FSI score, we also examine a sub-index focused on internal cohesion (specifically, the

FSI Cohesion indicators, which include measures of security apparatus, factionalized elites, and group

grievance). This allows us to see whether language attrition is specifically associated with the kinds of internal

divisions and grievances that could signal “Balkanisation.” As a robustness check, we also consider alternative

instability metrics: for instance, the Political Stability and Absence of Violence index from the World Bank’s

Worldwide Governance Indicators, and the incidence of internal conflict (e.g., number of violent conflict events

per year from datasets such as ACLED and UCDP). These alternatives help ensure our findings are not an artefact

of any single measurement approach.

Page 4532

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

Independent Variable: Arabic Attrition Index (AAI)

The key independent variable of interest is an index capturing Arabic language attrition in each country-year.

Guided by the SPH-LENS framework, we construct an Arabic Attrition Index (AAI) that quantifies the extent

to which Arabic is losing ground in various domains. This composite index is built from multiple indicators

reflecting the Socioeconomic (S), Political (P), and Historical (H) dimensions of language vitality. In practice,

assembling this index involves gathering data from a range of sources (see Annex Table A1 for a detailed list of

variables and data sources). For each country and year, we collate metrics such as:

 Education and Science: the percentage of secondary or tertiary education institutions where the primary

medium of instruction is Arabic vs. English/French (from UNESCO and national statistics), the share of

scientific publications or higher-degree theses published in Arabic (e.g. from bibliometric databases),

and the proportion of patent applications filed in Arabic (from WIPO data).

 Digital and Media: the fraction of web content or media output in Arabic. For example, the percentage

of websites with content in Arabic, and the volume of Arabic-language content on platforms like

Wikipedia or major social media.

 Language Use and Attitudes: survey-based measures of English proficiency (such as EF’s English

Proficiency Index, where rising proficiency may indicate shifts away from Arabic in daily use) and public

opinion surveys on language preference for education or work (when available).

 Official Status and Policy: whether Arabic is the sole official language or one among others, any changes

in constitutional language provisions (from sources like the Constitute Project), the presence of national

language academies or government programs for Arabic preservation, and state investment in Arabic-

language media and education. We also note any major language policy changes (such as introducing

English as a mandatory medium for certain subjects).

 Historical/Structural Factors: a dummy variable for countries with a colonial legacy of French or British

rule (since that often correlates with entrenched use of French/English among elites and institutions), and

an ethnolinguistic fractionalisation index (to account for the presence of sizable non-Arabic linguistic

groups within the country, which could affect Arabic’s role). While these factors change little over time,

they provide important context; in the panel analysis, country fixed effects will absorb purely time-

invariant factors like colonial history, but we explore interactions (e.g., whether language attrition has a

stronger effect on instability in ex-colonial states).

To address concerns about data reliability and aggregation, we adopt a deliberately conservative and transparent

strategy in constructing the AAI. All component indicators are first standardised to a common scale so that higher

values consistently capture greater erosion in Arabic’s societal role. We then aggregate them in two steps. Within

each of the Socioeconomic, Political and Historical dimensions, indicators are averaged after standardisation,

which prevents any single series from dominating its dimension purely because of scale differences. Across

dimensions, we give somewhat greater implicit weight to socioeconomic indicators (education, scientific output,

media and digital content), reflecting both their denser temporal coverage and their closer theoretical connection

to domain loss in high‑stakes arenas, while still preserving the contribution of political and historical factors.

Sensitivity checks using alternative schemes—such as equal weighting of all indicators irrespective of dimension

or principal‑component‑based weights—yield highly correlated AAI series and do not alter the main regression

results, suggesting that our substantive findings are not an artefact of a particular weighting choice.

Data limitations are unavoidable in a cross‑national, multi‑decade panel of this kind, particularly in domains

such as digital media or bibliometric series where coverage improves markedly over time and varies across

countries. We therefore adopt a set of simple rules to handle missing values. Short gaps in otherwise

well‑behaved time series (typically one to two years) are linearly interpolated, while longer gaps are left missing

so as not to fabricate artificial precision. For indicators that are structurally sparse (for example, early‑period

internet usage or web‑content measures), we rely more heavily on later years when measurement has stabilised,

and we down‑weight clearly noisy series in the composite index. As a robustness check, models estimated on a

Page 4533

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

reduced sample with only minimally imputed data produce coefficients for AAI that are very similar in

magnitude and significance to those in the full sample, indicating that our results are not driven by any particular

imputation choice.

Cross‑national comparability also merits caution. Some sources, such as national statistics on the language of

instruction or internal administrative reports on language policy, differ in detail and classification across states.

Wherever possible we harmonise categories ex post (for example, by distinguishing “primarily Arabic‑medium”

from “primarily foreign‑medium” provision rather than relying on finer national typologies) and focus on

within‑country changes over time rather than absolute levels. The use of country fixed effects further mitigates

concerns that persistent differences in measurement practices or institutional setups contaminate our estimates:

any time‑invariant biases in how countries record education, media or language policy are absorbed by these

fixed effects. Nonetheless, we acknowledge that the AAI remains a best‑effort proxy constructed from

heterogeneous data, and some degree of measurement error is inevitable. This reinforces our decision to interpret

the index as a broad structural signal of language attrition rather than as a finely calibrated measure of linguistic

behaviour.

Each indicator is normalized (scaled so that higher values consistently signify greater attrition risk or language

decline) and then combined into the AAI composite score for a given country-year. We assign weights to

components based on theoretical importance and data reliability, following the approach outlined in our SPH-

LENS framework (see Annex A for details on weighting and normalization schemes). Conceptually, a higher

AAI indicates a greater erosion of Arabic’s vitality (i.e., more domains where Arabic is diminishing), whereas a

lower AAI means Arabic remains relatively robust. By design, the index is intended as an early-warning metric:

significant movements in these indicators should signal risk well before Arabic is no longer passed to the next

generation. For instance, if the share of university courses taught in Arabic drops sharply or Arabic web content

plummets, such trends would raise the AAI even if virtually all children still learn Arabic at home – warning of

future attrition if unaddressed.

Control Variables

We incorporate a set of control variables to account for other factors that might influence state stability and could

correlate with language attrition. This is crucial for isolating the effect of language decline amid a complex socio-

political context. Key controls include:

 Economic development: GDP per capita (in constant USD, logged) to control for general development

level (wealthier countries tend to be more politically stable on average and more globalized, which could

both encourage English penetration and provide resources to mitigate conflict).

 Socioeconomic inequalities: measures of economic strain such as income inequality (Gini index) and

youth unemployment rate. High inequality or large pools of unemployed youth can fuel instability and

unrest and might also drive emigration or adoption of foreign languages among the disaffected populace.

 Demographics: the youth bulge (the percentage of young adults in the population). A large youth cohort

can strain job markets and social services, potentially contributing to unrest; it might also be more

inclined toward global cultural influences, including language shifts. We control for this to ensure our

language index isn’t inadvertently proxying a demographic effect.

 Education level: overall education attainment (e.g. adult literacy rate or average years of schooling).

Higher education levels can have mixed effects – they often promote stability via human development

but also tend to increase bilingualism and the use of English. Including education helps separate general

education effects from language-specific effects.

 Globalization and connectivity: urbanization rate and internet penetration. More urban, digitally

connected populations may be simultaneously more exposed to foreign languages (facilitating attrition)

and more capable of political mobilization (possibly affecting stability).

Page 4534

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

 Governance and institutions: measures of governance quality (e.g., government effectiveness or

corruption index from the Worldwide Governance Indicators) and political regime type or openness (such

as a Freedom House score or Polity index). Poor governance can cause instability (and might coincide

with poor education or language policy), whereas very authoritarian or very democratic regimes might

have different stability dynamics as well as different stances on language (for instance, some

authoritarian regimes actively promote Arabic nationalism, while others might neglect it). We also

employ country fixed effects (see below), which inherently control for any time-invariant country

characteristics – geography, historical cleavages (sectarian or ethnic divisions), etc. – that might influence

stability. This means, for example, that a country with a historically stronger pan-Arab identity or unique

linguistic situation will have that baseline accounted for, and our analysis will focus on within-country

changes over time.

Analytical Strategy

The paper employs a multi-pronged statistical analysis approach:

Panel Fixed-Effects Regression: Our main analysis uses panel regression models with country fixed effects.

The baseline specification regresses the Fragile States Index score on the lagged Arabic Attrition Index,

controlling for the aforementioned factors, and includes year fixed effects to absorb global or region-wide shocks

(e.g., worldwide economic crises or the 2011 Arab Spring). By using country fixed effects, we control for all

stable characteristics of countries, so the estimates leverage within-country, over-time variation. Essentially, we

ask: in years when a given country experiences a greater decline in Arabic (higher AAI), does it also see a

subsequent uptick in fragility, relative to its usual baseline level?

The model can be expressed as:

FragilityIndex_풊풕= 휷 ⋅ AttritionIndex_{풊,풕−ퟏ}+ 휸 ⋅ 푿_풊풕+ 휶_풊+ 휹_풕+ 휺_풊풕,

where α are country fixed effects and δ are year fixed effects, and 푿_풊풕represents the vector of control variables.

We lag the Attrition Index by one year (and test multi-year lags in some specifications) to reflect the expectation

that language shifts might precede and gradually contribute to instability, and to mitigate simultaneity concerns

(avoiding use of a contemporaneous value that could be jointly determined with instability). Standard errors are

clustered at the country level to account for serial correlation within each country’s time series. This fixed-effects

approach focuses on changes within each country, thereby factoring out cross-country differences in baseline

stability and linguistic environments.

Granger Causality Tests

To probe the direction of causality, we conduct panel Granger causality analyses. While our theoretical model

posits that language attrition leads to instability (i.e., loss of Arabic cohesion causes fragmentation), it is also

plausible that causality runs the other way (instability or conflict might disrupt the use of the standard language

or fragment education systems, thus accelerating attrition). The paper test both directions by estimating vector

autoregression (VAR) models in a panel context. Specifically, we examine whether past values of the AAI

significantly improve the prediction of current fragility (beyond the information provided by past fragility itself

and controls), and vice versa. In practice, this involves including multiple lags of AAI and FSI in a system of

equations and applying Wald tests for the joint significance of those lags. A finding that lagged language attrition

indicators have a significant effect on fragility, but not so much the reverse, would support the hypothesized

direction (language decline as a precursor to instability). We also inspect impulse response functions from the

panel VAR to illustrate the temporal dynamics – for example, whether a shock to the language index (a sudden

drop in Arabic usage) leads to a gradual rise in fragility over subsequent years.

Robustness Checks

We perform several robustness checks to validate the stability of our results. First, we estimate alternative models

such as a random-effects panel model and a first-differences model (which looks at year-to-year changes) to see

Page 4535

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

if the core findings persist under different assumptions. (A Hausman test is used to compare fixed vs. random

effects; we anticipate fixed effects is preferable given potential correlation between our language index and

unobserved country traits.) We also try including a lagged dependent variable (the prior year’s fragility score)

in the regression to account for the persistence of fragility; this is a more stringent test since it controls for

baseline stability levels – we employ a system GMM estimator in this case to address the Nickell bias that arises

from including a lagged dependent in a fixed-effects panel. Second, we experiment with different subsets of

indicators for the AAI (e.g., using only the socioeconomic indicators, or only the policy-related ones) to see if

any particular dimension is driving the results or if the combined index is robustly associated with fragility across

variations. Third, we conduct subgroup analyses: for example, splitting the sample between countries of the

Maghreb (Northwest Africa, with strong French influence) vs. the Mashreq (Eastern Arab countries), or between

wealthier Gulf states vs. lower-income states, to check if the relationship holds in each subgroup. This can reveal

if, say, oil-rich Gulf monarchies – which have extensive English use domestically but strong state capacity –

deviate from the pattern observed in other states. Fourth, we check for outliers by dropping one country at a time

(a jackknife approach) to ensure that a chronically conflict-ridden country (like Somalia) or a uniquely

multilingual country (like Lebanon) isn’t unduly skewing the results. Finally, as noted, we test alternative

outcome measures (e.g., using the “Cohesion” component of FSI specifically, or counts of internal conflict

events) to ensure that the core finding – a link between language attrition and instability – is not dependent on

how instability is measured.

RESULTS

The AAI exhibits a general upward trend across most Arab countries over 2000–2025, signalling worsening

language vitality, though trajectories vary.

Morocco and Tunisia, for instance, show marked increases in AAI during the 2000s and 2010s as French and

later English expanded in education and business. Gulf states such as the UAE and Qatar begin with relatively

high AAI scores because of entrenched English use in labour markets and universities; some show slight

improvement or stabilisation in the early 2020s following new language initiatives. Conflict-affected states (Iraq,

Syria, Yemen, and Somalia) display erratic patterns as wars disrupt standardised education and media, pushing

communication into dialects or other languages. Relatively conservative and stable countries such as Saudi

Arabia keep lower AAI scores, though even their gradual domain loss is visible in technology and higher

education.

FSI scores span a wide range, from relatively stable Gulf monarchies (FSI in the 30s) to highly fragile states like

Yemen, Somalia and Sudan (FSI above 100). The average regional FSI deteriorated markedly in the early 2010s

during the Arab uprisings and subsequent conflicts, with some countries recovering partially and others

continuing to worsen. These differences provide sufficient variation to test whether higher AAI tends to be

associated with higher fragility.

A simple bivariate plot (not shown here) reveals a positive correlation between AAI and FSI across country

years: high attrition observations tend to be high fragility observations. Annex Table A3 reports a Pearson

correlation of about +0.65 between AAI and FSI and +0.60 between AAI and the FSI Cohesion subindex,

indicating a substantial association even before controls are added.

Fixed-Effects Regression: AAI as Predictor of Fragility

The baseline fixed effects model (Model 1, Annex Table A4) regresses FSI on lagged AAI with country and

year fixed effects. The coefficient on AAI_(t−1) is positive (0.50) and highly significant (p < 0.001), implying

that a one-point increase in the AAI is associated with a 0.5-point increase in the FSI score the following year.

In the full model with controls (Model 2, Annex Table A5), the effect of AAI_(t−1) remains positive and

significant (coefficient ≈ 0.60, p = 0.001). Substantively, a one standard deviation increase in AAI corresponds

to roughly a 2.5-point rise in next year's FSI, holding other factors constant. This effect is comparable in size to

medium-scale shifts in economic or governance indicators and suggests that language attrition is a non-trivial

driver of fragility. Controls behave as expected: higher GDP per capita, literacy and government effectiveness

Page 4536

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

are associated with lower fragility, while a larger youth bulge, higher inequality and greater internet use are

associated with higher fragility.

Using the FSI Cohesion subindex as the outcome (Model 3, Annex Table A6) yields a similar pattern: the

coefficient on AAI_(t−1) is about 0.25 (p ≈ 0.004). Countries experiencing rising attrition tend to register

worsening scores on security apparatus, factionalised elites and group grievance, consistent with our hypothesis

that erosion of a shared language undermines internal cohesion.

A dynamic specification including lagged FSI (Model 4, Annex Table A7) shows that fragility is highly

persistent (FSI_(t−1) ≈ 0.55). Yet AAI_(t−1) still exerts a positive and statistically significant effect (coefficient

≈ 0.30, p ≈ 0.006), indicating that language attrition predicts changes in fragility even after accounting for the

country’s recent stability trajectory.

An alternative specification using a three-year cumulative AAI produces an even stronger association, suggesting

that sustained attrition has cumulative effects on stability.

Granger Causality and Robustness

Panel Granger causality tests support the interpretation that language decline tends to precede instability rather

than simply result from it. Lagged AAI jointly and significantly improves prediction of FSI beyond lagged FSI

alone; in contrast, lagged FSI has a weaker and often insignificant effect on current AAI once past AAI is

controlled. In a two-lag panel VAR, we reject the null that “AAI does not Granger cause FSI” at the 5% level,

while failing to reject or only weakly rejecting the reverse null.

Impulse response functions illustrate this asymmetry: a positive shock to AAI leads to a rise in FSI that peaks

around one to three years later before gradually fading, while a shock to FSI produces at most a small, short-

lived increase in AAI. This temporal pattern is consistent with language attrition acting as an early warning

indicator for future instability.

Robustness checks reinforce the core findings. Random effects and first difference models both yield positive,

significant coefficients for AAI. Results persist when dropping countries one at a time, and the attrition–fragility

link remains when high-conflict cases are excluded, with some evidence of even stronger effects in subsets such

as the Maghreb. Alternative outcome measures (FSI Cohesion, conflict events, and PSAV scores) also point to

a positive relationship between AAI and instability.

Component-wise analyses reveal that the Socioeconomic dimension of AAI (education, media and digital use)

has the strongest association with fragility, while the Political dimension (legal status, formal policy) is less

predictive, likely because most constitutions continue to affirm Arabic’s official status. Historical factors such

as colonial legacy are largely time invariant, but interactions suggest that the impact of recent attrition is more

pronounced in ex-French colonies, where Arabic’s structural position is already relatively weak.

A brief comparison between Tunisia and Jordan illustrates the pattern. Tunisia’s shift towards more English (and

French) in higher education and economic sectors coincided with a moderate rise in AAI and a deterioration of

FSI scores, especially on cohesion indicators. Jordan, which maintained a stronger institutional role for Arabic,

experienced much less instability despite facing economic pressures. Many differences exist between the two

countries, but the pair reflects the broader relationship detected in the panel analysis.

Overall, the results consistently support the hypothesis that Arabic language attrition is linked to higher levels

of state fragility and internal tension, and that this link is not an artefact of a small set of outliers or any model

specification.

Illustrative Country Cases

To make the aggregate patterns more concrete, we briefly consider how language attrition interacts with political,

economic and cultural conditions in specific country contexts. Tunisia and Jordan offer an instructive contrast.

In Tunisia, the post‑2000 period saw a gradual but clear expansion of French and English in higher education,

Page 4537

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

business and technology, reflected in a rising AAI. These changes occurred alongside—and were partly shaped

by—broader economic liberalisation, uneven regional development and contentious struggles over the post‑2011

political order. As fragility intensified, particularly on cohesion‑related indicators, public debates increasingly

framed language as one axis of a deeper divide between cosmopolitan, foreign‑language‑proficient elites and

more marginalised, predominantly Arabic‑speaking groups. In our framework, Tunisia exemplifies a scenario in

which language attrition amplifies existing structural tensions and helps to structure perceptions of inequality

and exclusion.

Jordan, by contrast, has experienced significant economic pressures, demographic strain from refugee inflows

and periodic episodes of protest, yet its AAI remains comparatively lower and more stable. Arabic has retained

a strong institutional presence in schooling, public media and official discourse, even as English has expanded

in certain sectors. Jordan’s fragility scores do fluctuate over the period, but they do not exhibit the same sustained

deterioration on cohesion indicators observed in Tunisia. We do not claim that stronger Arabic vitality explains

Jordan’s relative stability, which is also shaped by regime strategies, external support and security arrangements.

However, the comparison highlights a plausible mechanism through which language policies—by sustaining a

shared communicative and symbolic infrastructure—may help to moderate fragmentation in otherwise

challenging environments.

Similar dynamics appear in other cases. In parts of the Maghreb, for example, accelerated shifts toward French

and English in higher education and high‑status employment have coincided with debates over identity,

marginalisation and the role of Arabic in public life. In several Gulf states, entrenched English dominance in the

private sector and tertiary education coexists with efforts to reassert Arabic in official communication and

national branding; here, rising AAI values coexist with relatively lower fragility scores, suggesting that high-

capacity states may be better able to manage the tensions generated by sociolinguistic dualisation. Taken

together, these cases illustrate that language attrition operates through interaction with political and economic

structures rather than in isolation, and that its consequences depend on how states and societies respond to

emerging linguistic stratification.

Counterfactual Trajectories: Fragility Without Major Language Shift

Our panel also contains episodes in which states experienced heightened fragility without clear evidence of a

preceding, large‑scale shift away from Arabic in the domains we measure. In some conflict‑affected contexts,

for instance, sharp spikes in the FSI are driven by external interventions, regime collapses or localised power

struggles that unfold largely within an Arabic‑dominant linguistic environment. In these cases, the AAI either

changes only modestly or evolves on a different timetable from the instability shock. Such trajectories underscore

that language attrition is neither a necessary nor a sufficient condition for state fragility: states can become highly

unstable even when Arabic remains the principal medium of education, media and official discourse.

These counterfactual scenarios are analytically useful for two reasons. First, they demonstrate that the positive

association between AAI and fragility we document does not simply reflect a mechanical co‑movement of all

risk indicators but rather a pattern that coexists with notable exceptions. Second, they help to clarify the role of

language within a broader causal constellation. Where fragility rises in the absence of major language shift,

factors such as abrupt institutional breakdown, military intervention, resource shocks or deep‑seated sectarian

conflict appear to dominate the dynamics of instability. In contrast, the cases highlighted earlier suggest that,

when language attrition is pronounced and cumulative, it may interact with these other drivers by deepening

informational and symbolic divides between social groups, thereby making societies more susceptible to

polarisation and governance failure. Recognising both types of trajectories—fragility with and without major

language shift—allows for a more nuanced interpretation of the AAI as an early‑warning signal: it captures one

important dimension of structural risk but must be read alongside other political, economic and regional

indicators in any comprehensive assessment of state stability.

DISCUSSION

The findings provide quantitative support for a claim long present in Arab intellectual and policy debates: the

vitality of Arabic is not merely a cultural concern but a factor in national cohesion and security. This lends

Page 4538

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

empirical weight to the SPH framework, which treats language as an active variable embedded in socio-political

and historical structures, and to SPH LENS, which conceptualises language indicators as potential early warning

signals for instability.

The results also refine insights from Bourdieu and Gramsci. From a Bourdieusian perspective, the loss of

linguistic capital – as Arabic loses “market value” compared to English and French – undermines a key symbolic

resource binding citizens into a shared social order. From a Gramscian angle, the spread of hegemonic foreign

languages can entrench internal cultural cleavages, particularly between globally connected elites and less

connected populations, potentially sowing resentment and distrust. The fact that socioeconomic indicators of

attrition (education and media language) correlate more strongly with fragility than purely legal ones underscores

that what matters most is the language of everyday high-stakes interactions: schooling, knowledge production,

employment, and digital communication.

When those domains increasingly operate in a foreign language, segments of society may effectively inhabit

different linguistic and informational worlds. An urban, English-speaking elite may consume global media and

technocratic narratives, while a largely Arabic-speaking populace relies on different sources and discourses.

Such divides can weaken mutual understanding and make societies more vulnerable to polarisation and

mobilisation along identity lines. Our quantitative results provide a macro-level confirmation of dynamics that

have been reported qualitatively in several Arab contexts.

Policy Implications for Arab States and the Arab League

If language attrition is indeed associated with higher fragility, then language policy should be viewed as part of

national security strategy. For individual Arab states, this implies that ministries of education, culture,

information and defence need to coordinate in monitoring language trends and designing interventions. Our

results do not argue against learning English or French; they highlight the need for balanced bilingualism that

preserves Arabic’s primacy in public life and civic communication while equipping citizens to participate in

global networks.

Practically, governments could:



Monitor language vitality through a dedicated unit – for example, a Language Vitality Monitoring Unit

within a national security council – tasked with tracking indicators akin to AAI and issuing regular

assessments and alerts.

Design education policies that maintain Arabic as a core medium of instruction, especially in

foundational and civic subjects, while offering strong foreign language education. Where English or

French is introduced as a medium in scientific or technical fields, parallel investments in Arabic

terminology, textbooks and academic publishing can prevent a complete shift away from Arabic.



Invest in Arabic media and digital content, including support for high-quality Arabic content creation,

technology (search, translation, NLP) and pan-Arab cultural production that makes Arabic a language of

modernity and innovation, not only heritage.

At the regional level, the Arab League could sponsor an Arab Language Security Initiative that standardises

monitoring frameworks (drawing on SPH LENS), facilitates sharing of best practices, and coordinates

investments in Arabic language education and media. Such an initiative could parallel, in some respects, the role

of the Organisation Internationale de la Francophonie for French, albeit tailored to the specific historical and

religious significance of Arabic.

Information, Ideology and Resilience

The language–security link also intersects with information security and ideological resilience. A strong national

language provides a shared channel through which governments, civil society and media can communicate with

citizens and build common narratives. When public discourse fragments across languages, external actors – from

Page 4539

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

foreign states to transnational corporations or extremist groups – may find it easier to target segments with

tailored content. Arabic-language information ecosystems that are robust, diverse and credible may help

inoculate societies against both external propaganda and internal sectarian or extremist messaging.

Limitations and Future Refinements

While the Granger causality tests and dynamic panel models are consistent with the interpretation that increases

in Arabic attrition tend to precede and predict subsequent rises in fragility, they do not by themselves establish

a fully identified causal mechanism. Granger precedence shows that past movements in the AAI improve the

prediction of future fragility beyond past fragility alone, but it cannot rule out the influence of omitted variables

that evolve on similar time scales. Regional geopolitical shocks, slow‑moving ideological shifts, or changes in

global economic integration, for instance, could plausibly affect both language practices and state stability in

ways that are only imperfectly captured by our control variables.

Moreover, the sociopolitical processes at stake are complex and likely involve feedback loops. Episodes of

instability may erode the institutional environments that sustain Modern Standard Arabic in education and media,

even as longer‑term language stratification helps to structure patterns of grievance and elite–mass distance. Our

design, which relies on annual national‑level data, is better suited to detecting broad temporal associations than

to disentangling these finer, potentially bidirectional mechanisms. For these reasons, we interpret the positive

and robust AAI coefficients as evidence of a strong language–fragility nexus at the structural level, not as proof

that language attrition alone mechanically “causes” instability. Future work using subnational data, natural

experiments around major language‑of‑instruction reforms, or micro‑level survey and behavioural evidence will

be needed to more tightly identify the causal pathways suggested by our findings

Several limitations warrant caution. First, our AAI is an innovative but imperfect proxy for language vitality,

relying on available quantitative indicators that may miss qualitative nuances such as depth of proficiency or

attitudes toward language. More fine-grained survey data and better statistics on language use in education,

media and online platforms would allow a more precise index. Second, our design is observational; while panel

methods and timing tests support a causal interpretation, unobserved factors (e.g., cultural globalisation, shifting

regional alliances) could partly drive both language and stability. Future work could exploit natural experiments,

such as abrupt language of instruction reforms or subnational variation, to sharpen causal inference.

Third, our focus on the Arab world enhances internal comparability but limits external generalisation. It remains

to be seen whether similar patterns exist elsewhere, for example, in the decline of national or indigenous

languages vis-à-vis English in parts of Africa or Asia, or in multilingual states like India. The specific religious,

historical and diglossic features of Arabic may amplify the language–security link in ways that do not fully

translate to other contexts.

CONCLUSION

This study provides, to our knowledge, the first large-N quantitative test of the proposition that Arabic language

attrition is linked to national fragility. Building on the SPH and SPH LENS frameworks, we constructed an

Arabic Attrition Index for the 22 Arab League states over roughly 2000–2025 and examined its relationship with

the Fragile States Index and related measures of instability. Using panel fixed effects regressions, Granger

causality tests and multiple robustness checks, we found that increases in language attrition systematically

precede and predict higher levels of state fragility and internal cohesion problems, even after accounting for

economic, demographic and institutional factors.

Theoretically, these findings validate key elements of SPH and SPH LENS, demonstrating that language

indicators can help explain and anticipate political outcomes. Empirically, they introduce a composite index and

dataset that future researchers can refine and apply, whether to deepen analysis within the Arab region or to

compare across regions. Practically, they support a reframing of Arabic language policy as a strategic domain of

governance, with implications for education, media, technology and regional cooperation.

Page 4540

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

Future research should move in three directions. First, micro-level studies could examine how individual

bilingualism, language preferences and media consumption patterns relate to political attitudes, social trust and

mobilisation. Second, subnational and historical analyses could exploit regional variation and policy shocks

within states to better identify causal mechanisms. Third, comparative work could explore whether similar

dynamics arise where other lingua francas or national languages lose ground to global languages.

Ultimately, our central message is straightforward: language matters for the stability of states. In the Arab world,

Arabic is not only a vehicle of communication but also a shared symbolic and cultural infrastructure. Its gradual

marginalisation in crucial domains is unlikely to be neutral. Conversely, strengthening Arabic – through

thoughtful, balanced policies – can be seen as an investment in resilience. We therefore urge policymakers to

move beyond a purely cultural framing and to recognise language vitality as an integral component of national

security planning. Safeguarding Arabic is, in this sense, part of safeguarding the future cohesion and stability of

Arab societies.

REFERENCES

1. Eberhard, David M., Gary F. Simons, and Charles D. Fennig, eds. 2021. Ethnologue: Languages of the

World. 24th ed. Dallas, TX: SIL International.

2. Fishman, Joshua A. 1991. Reversing Language Shift: Theoretical and Empirical Foundations of

Assistance to Threatened Languages. Clevedon: Multilingual Matters.

3. Bourdieu, Pierre. 1991. Language and Symbolic Power. Cambridge, MA: Harvard University Press.

4. Gramsci, Antonio. 1971. Selections from the Prison Notebooks. Edited and translated by Quintin Hoare

and Geoffrey Nowell Smith. New York: International Publishers.

5. Stanley Dubinsky, Harvey Starr, Weaponizing Language: Linguistic Vectors of Ethnic

Oppression, Global

Studies

Quarterly,

Volume

2,

Issue

2,

April

2022,

ksab051, https://doi.org/10.1093/isagsq/ksab051

6. Suleiman, Yasir. 2003. Arabic Language and National Identity: A Study in Ideology. Edinburgh:

Edinburgh University Press.

7. SULEIMAN, YASIR. The Arabic Language and National Identity: A Study in Ideology. Edinburgh

University Press, 2003. http://www.jstor.org/stable/10.3366/j.ctvxcrx0x .

8. Mustafa Taha, “Arabizi: Is Code-Switching a Threat to the Arabic Language,” in The Asian Conference

on Arts & Humanities 2015: Official Conference Proceedings (Osaka: The International Academic

Forum (IAFOR), 2015).

9. UNESCO Ad Hoc Expert Group on Endangered Languages. 2003. Language Vitality and Endangerment.

Document submitted to the International Expert Meeting on UNESCO Programme Safeguarding of

Endangered

Languages,

Paris,

10–12

March

2003.

Paris:

UNESCO.

https://ich.unesco.org/doc/src/00120-EN.pdf .

10. Anderson, Benedict. 1983. Imagined Communities: Reflections on the Origin and Spread of Nationalism.

London: Verso.

11. Gellner, Ernest. 1983. Nations and Nationalism. Ithaca, NY: Cornell University Press.

12. Suleiman, Yasir. 2003. Arabic Language and National Identity: A Study in Ideology. Edinburgh:

Edinburgh University Press.

13. Greenberg, Robert D. 2011. “Language, Identity, and Balkan Politics: Struggle for Identity in the Former

Yugoslavia.” Occasional Paper No. 216. Washington, DC: Woodrow Wilson International Center for

Scholars.

14. Marshall, Monty G., and Gabrielle Elzinga-Marshall. 2017. Global Report 2017: Conflict, Governance

and State Fragility. Vienna, VA: Center for Systemic Peace.

15. Fearon, James D., and David D. Laitin. 2003. “Ethnicity, Insurgency, and Civil War.” American Political

Science Review 97 (1): 75–90.

16. Marson, Margherita, Mario Migheli, and Daniele Saccone. 2021. “New Evidence on the Link between

Ethnic Fractionalization and Economic Freedom.” Economics of Governance 22 (3): 257–292.

Page 4541

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

17. Shadi, Mohamed, and Mostafa Ahmed. 2025. “Arabs without Arabic: A Spatio‑Temporal Model of

Language Attrition and Its Implications for National Security.” International Journal of Research and

Innovation in Social Science 9 (10): 7167–7180. https://doi.org/10.47772/IJRISS.2025.910000584

18. UNESCO. 2010. Atlas of the World’s Languages in Danger. 3rd ed. Edited by Christopher Moseley.

Paris: UNESCO Publishing.

19. Fishman, Joshua A. 1991. Reversing Language Shift: Theoretical and Empirical Foundations of

Assistance to Threatened Languages. Clevedon: Multilingual Matters.

20. Shadi, Mohamed, and Mostafa Ahmed. 2025. “Arabs without Arabic: A Spatio‑Temporal Model of

Language Attrition and Its Implications for National Security.” International Journal of Research and

Innovation in Social Science 9 (10): 7167–7180. https://doi.org/10.47772/IJRISS.2025.910000584

21. Marshall, Monty G., and Gabrielle Elzinga-Marshall. 2017. Global Report 2017: Conflict, Governance

and State Fragility. Vienna, VA: Center for Systemic Peace.

ANNEX. Variable Definitions and Sources

Table A1. Variable Definitions and Sources

Variable

Fragile

Definition / Measurement

Composite annual score

Source (indicative)

state Fund for Peace, Fragile

(0–120)

indicating

States Index fragility/vulnerability (higher = more fragile). Aggregates 12 States Index reports

(FSI)

political, social, and economic indicators (cohesion, economic,

political, social). We use overall score for each Arab League state.

FSI

“Cohesion”

sub-index

Sub-component of FSI capturing internal cohesion and security Fund for Peace, Fragile

pressures. Average of: Security Apparatus, Factionalized Elites, States Index (Cohesion

Group Grievance. Approx. range 0–30 (higher = more internal category)

division). Used as alternative dependent variable.

Arabic

Composite index (constructed by authors) quantifying erosion of Authors’

calculations

Attrition

Index (AAI)

Arabic language vitality in each country-year. Built from based

Socioeconomic (S), Political (P), Historical (H) indicators per OpenAlex, W3Techs, EF

SPH-LENS. Higher AAI = more domain loss and weaker Arabic EPI, Constitute Project,

on

UNESCO,

status. Main components: (1) education language (share of World

Bank,

etc.

secondary/tertiary teaching in Arabic vs foreign languages), (2) (operationalising

share of scientific output and patents in Arabic, (3) Arabic share SPH-LENS framework)

of national web/media content, (4) English proficiency and usage,

(5) constitutional and policy status of Arabic, (6) colonial legacy

and structural factors.

GDP

capita

per Gross domestic product per capita (constant US dollars). Used in World

log form in regressions. Controls for level of development Development

(wealthier countries tend to be more stable). (WDI)

Youth bulge Share of population aged 15–29 years. Captures demographic UN World

pressure from a large youth cohort; often associated with higher Prospects; World Bank

risk of unrest if employment and integration are weak. population by age

Bank,

World

Indicators

Population

(%

population

15–29)

Adult

Percentage of population aged 15+ who can read and write. Proxy UNESCO Institute for

literacy rate for human capital and general education; helps isolate language- Statistics; World Bank

(%)

specific effects from overall education level.

Gini

index Income inequality index (0–100). Higher values = more World Bank, WDI (Gini

(inequality)

inequality. Controls for socio-economic stress which can drive index)

grievances and instability. (For missing years, nearest available

values or linearly interpolated estimates are used.)

Urbanisation

(%)

Urban population as a share of total population. Proxy for World Bank, WDI (Urban

modernisation and exposure to global culture. Effects on stability population %)

can

development and mobilisation capacity).

www.rsisinternational.org

be

ambiguous

(urbanisation

often correlates

with

Page 4542

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

Internet

Individuals using the internet, percent of population. Captures International

users (% of digital connectivity and exposure to online (often English- Telecommunication

population)

dominated)

mobilisation and networked protest.

Governance quality index (World Bank WGI). Approximate World Bank, Worldwide

information

environment;

also

relevant

for Union (ITU); World Bank

Government

effectiveness range –2.5 (weak) to +2.5 (strong). Measures perceived quality of Governance Indicators –

public services, policy formation and implementation. Better “Government

governance typically reduces fragility.

Government expenditure on education as share of GDP. Proxy for UNESCO Institute for

spending (% state commitment to the education system (including potential Statistics; World Bank

of GDP) investment in Arabic curricula, teacher training, etc.). education data

effectiveness”

Education

Panel Fixed-Effects Regression Equation: The main estimation is a fixed-effects panel model with lagged

language attrition and controls:

퐹푆퐼_푖,푡ꢀ = ꢀ훽₁ꢁ퐴퐴퐼_푖,푡−1ꢀ + ꢀ훽₂ꢁ푋_푖,푡ꢀ + ꢀ훼_푖ꢀ + ꢀ훿_푡ꢀ + ꢀ휖_푖,푡,

⏟

Fragility

where 퐹푆퐼_푖,푡is the Fragile States Index for country i in year t; 퐴퐴퐼_푖,푡−1is the Arabic Attrition Index in the prior

year; 푋_푖,푡is a vector of control variables (GDPpc, youth bulge, literacy, etc. as defined above) for country i, year

t; 훼_푖are country fixed effects (absorbing time-invariant country traits); 훿_푡are year fixed effects (capturing shocks

common to all countries in year t); and 휖_푖,푡is the error term. All models are estimated on an annual panel (≈2000–

2023) of 22 Arab League countries with heteroskedasticity-robust standard errors.

Table

A2.

Summary

Statistics

(Panel

Sample,

22

Countries

~2000–2023)

Descriptive statistics for all variables (country-year observations ≈374). Mean and standard deviation are

calculated across the full panel sample, with minimum and maximum country-year values in parentheses.

Variable

Mean Std. Dev. Min

Max

111.3

30.0

90.0

60,000

35.0

98.0

45.0

100.0

99.0

1.20

8.0

N (country-years)

Fragile States Index (FSI)

FSI Cohesion sub-index

78.2

19.5

50.0

11,000

30.0

75.4

35.0

70.2

59.9

-0.50

4.0

22.3

7.8

34.7

8.0

374

350

374

300

Arabic Attrition Index (AAI)

GDP per capita (constant US$)

Youth bulge (% pop 15–29)

Adult literacy rate (%)

20.0

15,000

3.5

20.0

300

25.0

50.0

30.0

29.0

10.0

-2.50

1.5

15.2

5.0

Gini index (inequality)

Urbanisation (% of population)

Internet users (% of population)

Government effectiveness (WGI)

Education spending (% of GDP)

20.5

29.8

0.95

2.0

Note: N is the number of countries–year observations for which data are available. Some variables (e.g. Gini,

education spending) have fewer observations due to data gaps.

Page 4543

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

Table A3. Pairwise Correlations Among Key Variables

Lower-triangular matrix of Pearson correlation coefficients for all variables in the study. Each cell shows the

correlation between the row variable and column variable. (For example, the correlation between AAI and FSI

is +0.65, indicating higher language attrition is associated with higher fragility. Country and year fixed effects

are not included in these bivariate correlations.)

FSI

Cohes AAI GDP

Yout

h

Litera Gini

cy

Urban

Inter

net

GovE

ff

EduSp

end

ion

pc

1.00

0.90

0.65

FSI

(Fragility)

1.00

0.60

Cohesion

Sub-Index

1.00

Arabic

Attrition

Index

–0.50

0.40

–0.45

0.35

–

0.30

1.00

–0.60

0.70

–0.20

0.80

0.85

0.80

0.30

GDP

capita

per

0.10

0.30

0.20

0.30

0.40

1.00

Youth Bulge

(%)

–0.40

0.30

–0.35

0.25

–0.70

0.30

1.00

Adult

Literacy(%)

–0.40

0.60

0.70

0.50

0.40

1.00

Gini

(Inequality)

–0.40

–0.30

–0.80

–0.30

–0.35

–0.25

–0.75

–0.25

–0.50

–0.60

–0.70

–0.40

–0.30

–0.40

–0.60

–0.20

1.00

0.80

0.50

0.20

Urbanization

(%)

1.00

0.70

0.30

Internet Use

(%)

–

0.20

1.00

0.40

Gov.

Effectiveness

0.15

1.00

Education

Spending

Note: All correlations calculated over 22 countries × time. Bolded variables correspond to those used in

regression models. Cohesion is a component of FSI, hence a very high correlation. GDPpc = GDP per capita;

Gov. Effectiveness = Government Effectiveness (WGI). Correlations ≥0.30 in absolute value are significant at

p<0.05 (two-tailed).

Table A4. Fixed-Effects Panel Regression – Baseline Model (Dependent variable: Fragile States Index)

Model 1: Baseline fixed-effects regression with lagged Arabic Attrition Index as sole predictor (country and year

FE included). This tests the bivariate relationship between language attrition and next-year fragility, controlling

for unobserved country-specific factors and common yearly shocks.

Page 4544

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

Regressor Coefficient Std. Error t-statistic p-value

AAI (t−1)

0.50

0.12

1.05

4.17

0.000

Constant

57.30

54.60

Model statistics:



Observations (country–years): 374

Number of countries: 22

R-squared (within): 0.32

Country fixed effects: Yes

Year fixed effects: Yes

A one-point increase in the Arabic Attrition Index in year t−1 is associated with a 0.50-point increase in the FSI

score in year t, on average, controlling for country and year fixed effects

Table A5. Fixed-Effects Panel Regression – Full Model with Controls (Dependent variable: Fragile States Index)

Model 2: Fixed effects regression including the lagged Arabic Attrition Index and all control variables. This is

the primary specification testing language attrition’s effect on fragility while holding constant key economic,

demographic, and institutional factors.

Regressor

Coefficient Std. Error t-statistic p-value

AAI (t−1)

0.60

-0.15

0.10

-0.05

0.10

0.02

0.03

-4.00

0.18

0.05

0.04

0.02

0.04

0.03

0.01

1.00

0.11

5.50

3.33

0.001

0.003

0.025

0.015

0.013

0.506

0.036

0.000

0.070

0.000

GDP per capita (log)

Youth bulge (%)

Adult literacy (%)

Gini index

-3.00

2.25

-2.45

2.50

Urbanisation (%)

Internet users (%)

Government effectiveness

0.67

2.10

-4.00

-1.82

11.29

Education spending (% GDP) -0.20

Constant

62.10

Model statistics:



Observations: 340

Countries: 22

R-squared (within): 0.51

Page 4545

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025



Country fixed effects: Yes

Year fixed effects: Yes

After controlling for economic, demographic, and governance variables, lagged AAI retains a positive and highly

significant effect on FSI. The sign and magnitude of the controls correspond with expectations (e.g. higher GDP

per capita and better governance reduce fragility; larger youth bulge and higher inequality increase it).

Table A6. Fixed-Effects Regression – Cohesion Sub-Index as Dependent Variable

Model 3: Testing the impact of language attrition on the FSI Cohesion sub-index (internal cohesion and unity).

This uses the same controls as Model 2, but the dependent variable is the FSI Cohesion score (higher = more

internal division).

Regressor

Coefficient Std. Error t-statistic p-value

AAI (t−1)

0.25

-0.05

0.07

-0.03

0.08

0.01

0.02

-1.50

0.08

0.03

0.02

0.03

0.02

0.01

0.55

0.06

2.80

2.95

-1.67

2.33

-1.50

2.56

0.50

1.60

-2.73

-1.67

5.07

0.004

0.097

0.021

0.135

0.012

0.619

0.112

0.007

0.096

0.000

GDP per capita (log)

Youth bulge (%)

Adult literacy (%)

Gini index

Urbanisation (%)

Internet users (%)

Government effectiveness

Education spending (% GDP) -0.10

Constant

14.20

Model statistics:



Observations: 340

Countries: 22

R-squared (within): 0.44

Country fixed effects: Yes

Year fixed effects: Yes

Higher AAI in the previous year is significantly associated with higher cohesion-related fragility (greater

factionalism and group grievance). This supports the idea that erosion of a shared language undermines internal

cohesion.

Table

A7.

Dynamic

Panel

Model

–

Including

Lagged

Dependent

Variable

Model 4: Fixed effects regression for FSI including lagged FSI as an additional regressor (dynamic

specification). This tests robustness: whether AAI still predicts changes in fragility even after accounting for the

country’s prior fragility level.

Page 4546

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)

ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025

Regressor

Coefficient Std. Error t-statistic p-value

AAI (t−1)

0.30

0.55

-0.10

0.08

-0.03

0.05

0.01

0.02

-2.00

0.10

0.07

0.08

0.06

0.03

0.06

0.02

0.01

1.20

0.08

6.00

2.88

7.86

-1.25

1.33

-1.00

0.83

0.50

1.65

-1.67

-0.63

4.42

0.006

0.000

0.213

0.185

0.320

0.408

0.620

0.101

0.096

0.531

0.000

FSI (t−1)

GDP per capita (log)

Youth bulge (%)

Adult literacy (%)

Gini index

Urbanisation (%)

Internet users (%)

Government effectiveness

Education spending (% GDP) -0.05

Constant

26.50

Model statistics:



Observations: 318

Countries: 22

R-squared (within): 0.67

Country fixed effects: Yes

Year fixed effects: Yes

Fragility is persistent over time (lagged FSI coefficient ≈ 0.55), but lagged AAI remains a statistically significant

predictor of FSI even after controlling for the previous year’s fragility. This supports the interpretation that

language attrition tends to precede and contribute to increases in fragility rather than merely reflecting it.

Page 4547

www.rsisinternational.org