…Three-Year Audit Of 42 Attacks Exposes Accountability Gap In Security Response
1,369 Killed, 12 Arrests, No Convictions As Official Assurances Outpaced Justice
An analysis of official statements issued after major attacks from January 2023 to June 2026 reveals whether condemnatory rhetoric translated into arrests, prosecutions, and convictions.
How the Nigerian Government Has Recycled the Same Language Across 42 Security Crises, While 1,369 Nigerians Died and Not a Single Perpetrator Was Convicted
KEY FINDINGS AT A GLANCE
| 706× “Unfortunate” used Most dominant word — appears in 100% of responses regardless of death toll |
0 Convictions secured Across all 42 incidents, 1,369 deaths, and 389 pledges to ‘bring to book’ |
108:1 Bandit ratio Condemnation phrases to actual arrests — worst ratio of any blame category |
22% Copy-paste rate Of all 861 statement pairs, 191 share over 70% semantic similarity |
The Word That Buries the Dead
On Christmas Day 2023, gunmen descended on communities across Plateau State and killed 162 people. By the following morning, the Nigerian Presidency had issued its response. It was, the statement said, an “unfortunate” attack. Security forces had been deployed. The perpetrators would be “fished out“ and “brought to book.”
No one was ever convicted.
Three weeks earlier, 14 people had been killed in a Boko Haram attack in Borno State. The government’s response described it as an “unfortunate” attack. Security forces had been deployed. The perpetrators would be fished out and brought to book.
No one was ever convicted.
Four months before that, 87 schoolchildren were abducted from Niger State. The government condemned the “unfortunate” incident. Security forces were mobilized. Those responsible would be brought to book.
No one was ever convicted.
In 42 documented security crises since January 2023, Nigeria’s government used the word “unfortunate” 706 times. The number of convictions secured: zero.
This is not coincidence. It is a system.
A six-month investigation by The Whistler — analyzing 187 press statements from the Presidency, the National Security Adviser, the Ministry of Defense, the Nigerian Army, the Nigeria Police Force, NAN and major national newspapers across 42 documented security incidents, has found that Nigerian government responses to mass killings, kidnappings, and terrorist attacks follow a rigid, recycled four-part template that has remained essentially unchanged since January 2023, regardless of the scale of the atrocity, the region affected, or the identity of the perpetrators.
The investigation further found that the words most deployed in these statements are “condemn,” “dastardly,” “heinous,” and “brought to book”. They bear no statistical relationship to any subsequent law enforcement action. They are, in measurable terms, empty.
Part I: The Rhetoric — What the Government Says
Across the 42 incidents analyzed, spanning bandit attacks, Boko Haram and ISWAP strikes, herder/farmer clashes, and mass kidnappings, The Whistler identified a consistent set of rhetorical phrases that appear in virtually every government response.
The findings, illustrated in Figure 1, reveal a clear hierarchy of preferred language:
• “Unfortunate” — 706 occurrences. Used in every single response across all 42 incidents.
• “Being investigated” — 258 occurrences. Invoked in 142% of articles (averaging more than once per article).
• “Fish out” / “fished out” — 212 occurrences. A promise of pursuit that The Whistler found led to arrest in just 5 of 42 incidents.
• “Brought to book” — pledged 389 times. Resulting in zero convictions.
• “Dastardly” — used 229 times. “Heinous” — 165 times. Neither word has ever appeared in a subsequent charge sheet.
Figure 1: Government rhetoric keyword frequency across 187 press statements, January 2023 – June 2026.
What is most telling is not the frequency of these words individually, but their relationship to each other. “Unfortunate” — a word ordinarily reserved for minor mishaps — appears at nearly three times the rate of “being investigated.” The government is three times more likely to express that something is sad than to promise it will be looked into.
“Unfortunate” is not an expression of grief. It is a format. A rhetorical escape hatch that acknowledges a crisis without committing to any response to it.
Part II: The Template — The Copy-and-Paste Architecture of Impunity
To establish whether these phrases were part of a coordinated communication pattern, The Whistler applied TF-IDF (term frequency–inverse document frequency) vectorization and cosine similarity analysis across all 42 unique press statements. The results were unambiguous.
Of 861 unique statement pairings tested, 191 — 22 percent — returned a similarity score above 70 percent. Multiple pairings returned scores of 100 percent, meaning the statements were structurally identical save for the substitution of a location name and a casualty figure.
Figure 5: Semantic similarity analysis across 42 government press statements. 191 of 861 pairs exceed 70% cosine similarity. Four template families account for 100% of statements.
The analysis identified four distinct template families — labelled A, B, C and D — that account for the entirety of government responses across the three-year period. Each template follows the same four-step structure:
• Step 1 — Minimize: Apply the word “unfortunate” and express sorrow. No official is named. No institution is indicted.
• Step 2 — Promise action (passive voice): Security forces “have been deployed.” Perpetrators “will be fished out.” The passive construction ensures no officer, unit or commander is accountable for delivery.
• Step 3 — Signal competence: The government is “on top of the situation.” This phrase appears 168 times, typically within 24 hours of an incident, regardless of operational reality.
• Step 4 — Close with commitment (no deadline): The government is “committed to addressing the security challenges.” No timeline. No metric. No name.
The practical consequence is that a statement about the Christmas Day massacre of 162 civilians in Plateau State shares its complete syntactic skeleton with a statement about the killing of 7 soldiers in an ISWAP ambush in Borno State. Scale of atrocity has zero effect on depth of response. The template absorbs the crisis and emits a signal of governance without requiring any governance.
The template system is not a communication failure. It is institutional architecture designed to consume accountability.
Part III: The Action Gap — What the Government Does
The most revealing metric in this investigation is the condemnation-to-action ratio: the number of strong condemnation phrases used in government responses, divided by the number of concrete accountability outcomes (arrests plus prosecutions) that followed.
Figure 4: The accountability funnel — 42 incidents, 42 condemnations, 5 arrests, 4 prosecutions, 0 convictions.
Across 42 incidents, involving 1,369 confirmed deaths and 257 abductions:
• Only 5 incidents — 12 percent — led to any arrest at all.
• Only 4 incidents — 10 percent — resulted in any criminal prosecution.
• Zero incidents — 0 percent — ended in a conviction.
• Only 1 incident — the Christmas 2023 Plateau massacre with 162 deaths — triggered any documented policy shift.
Meanwhile, the government used the phrase “brought to book” 389 times. It made 12 arrests across the entire period. It secured no convictions.
Figure 3: Monthly frequency of “unfortunate” in government statements versus confirmed deaths, January 2023 – June 2026. No correlation exists between death toll and response depth.
Part IV: The Label — How a Word Determines Whether Anyone Is Held Accountable
Perhaps the most consequential finding of this investigation is the relationship between how perpetrators are labelled and whether any legal action follows. The Whistler’s analysis found that the choice of blame label is not merely semantic, it is structurally determinative of whether any law is invoked at all.
Figure 2: Condemnation-to-action ratio by perpetrator label. “Bandits” — 14 incidents, 444 deaths — ratio of 108:1 with zero arrests and zero prosecutions.
The data shows three distinct tiers of state response, determined entirely by the word used to name the attackers:
“UNKNOWN GUNMEN”
Applied exclusively to herder-farmer clashes, predominantly in South East, Plateau, Kaduna and Niger States, the label “unknown gunmen” is the most evasive designation in the government’s vocabulary. With 13 incidents and 811 confirmed deaths, the highest fatality total of any category. It generates a condemnation-to-action ratio of 9.1:1. Crucially, the label serves a political function: it avoids naming any ethnic or pastoral community, insulating the government from accusations of partiality in the Fulani and Biafra question while simultaneously mapping to no specific statutory instrument.
“BANDITS”
This is the investigation’s most significant finding. In 14 incidents in which attackers were labelled “bandits” — principally in Zamfara, Katsina and Kaduna States, with 444 confirmed deaths. The government issued 108 strong condemnation phrases. The number of arrests: zero. The number of prosecutions: zero.
The reason is legal. The Terrorism (Prevention) Act 2011, as amended, mandates federal prosecution, asset freezes, travel bans, and engagement of international law enforcement including INTERPOL. By classifying attackers as “bandits” rather than “terrorists,” the government explicitly avoids triggering this framework. The word “bandit” keeps the matter within the ordinary criminal code where capacity is lower, coordination is absent, and political will is thinner.
Calling them “bandits” instead of “terrorists” is not a classification error. It is a legal decision that closes the door on every enforcement mechanism the state would otherwise be required to activate.
“TERRORISTS”
Even when the government applies the label “terrorist”, as it does for Boko Haram and ISWAP attacks; the condemnation-to-action ratio remains a damning 23.3:1. Nine incidents. 114 deaths. Three arrests. Zero prosecutions. The label is used, but the legal apparatus it implies is not engaged.
Figure 6: Blame vocabulary divergence — rhetoric intensity vs accountability outcomes across all four blame categories.
The Grammar of Impunity
The Nigerian government has, since at least January 2023, operated a sophisticated communications system that performs accountability without delivering it. It is not that officials fail to notice the pattern, it is that the pattern serves a function.
When 162 people are killed on Christmas Day, the institutional response is a paragraph generated from a template, published within 24 hours, deploring the “unfortunate” and “dastardly” attack, promising that those responsible will be “brought to book,” and then, in the vast majority of cases — doing nothing further.
The language is not accidental. “Unfortunate” frames mass death as misfortune, not policy failure. “Bandits” prevents the Terrorism Act from being triggered. “Unknown gunmen” neutralizes the political charge of ethnic or pastoral accountability. “Brought to book” invokes the promise of legal consequence without activating any legal mechanism.
This investigation is not an argument that Nigerian security agencies are doing nothing. It is a precise and evidenced argument that the gap between what the government says and what it does is not a communications failure, rather it is the architecture of institutional apathy, embedded in language and reproduced across every category of crisis and every region of the country.
The word count of condemnation has no correlation with the body count of accountability. That is not a coincidence. It is a policy.
389 pledges to bring perpetrators to book. 12 arrests. 0 convictions. This is not a communication failure. It is the grammar of impunity.
METHODOLOGY & DATA SOURCES
This investigation is based on the collection and analysis of 187 press statements and news reports published between January 2023 and June 2026 across five primary sources: Punch Newspapers, Vanguard, ThisDay, the News Agency of Nigeria (NAN), and official Presidency/Aso Rock press releases. Statements were collected using a custom Python web scraper targeting government response language across all documented security incidents.
Keyword frequency analysis was performed using regular expression pattern matching on extracted government response text. Semantic similarity analysis was conducted using TF-IDF (term frequency–inverse document frequency) vectorization with 2–4-gram features, and cosine similarity computed across all 861 unique statement pairings using scikit-learn (Python 3).
Accountability outcomes — arrests, prosecutions, convictions and policy shifts — were cross-referenced against the Armed Conflict Location and Event Data (ACLED) Nigeria dataset, the Nigeria Security Tracker maintained by the Council on Foreign Relations (CFR), and reports by the National Human Rights Commission (NHRC). Incident fatality data is drawn from these sources; press statement data is drawn from the scraped corpus.
The condemnation-to-action ratio is defined as: total strong condemnation phrases used across all statements for a given incident ÷ (total arrests + total prosecutions following that incident). A ratio of infinity is assigned where the denominator is zero; in tables this is represented as the maximum observed value (11.0) for presentation purposes.
(The Whistler)