Keir Starmer announced his resignation as Prime Minister on 22 June 2026, and the speech he delivered from Downing Street that morning was, on close reading, the most first-person-singular-heavy speech of his premiership. "Every decision I've taken has been about putting the country I love first. That is why I will resign as leader of the Labour Party. I have spoken to His Majesty the King this morning to inform him of my decision. I will remain in post as Prime Minister until the contest is complete. I will do everything I can to ensure an orderly handover of power. I will also give my successor my full and unequivocal support." In roughly ninety seconds of prepared remarks he used the first-person singular pronoun nine times. The word we appeared once. The word our appeared twice, in the phrase our country. This was not accidental. It was the register the framework predicts a departing PM's authority-loss produces.
The framework being applied here is the I-We Index — a pronoun-frequency analysis grounded in James Pennebaker's forty-year body of work at the University of Texas on linguistic markers of leadership. Its central claim, well-supported in the academic literature and previously applied by this publication to CEOs of S&P 500 companies, is that leaders whose speech skews toward first-person-plural pronouns (we, our, us) tend to run organisations that perform better on retention, trust, coherence and crisis response, while leaders whose speech skews toward first-person-singular (I, my, me) — particularly under stress — display the pattern the literature calls narcissistic linguistic dominance, which correlates with organisational fragility. Applied to British Prime Ministers from Blair to Starmer, the pattern is observable, mechanically trackable from public speeches and PMQs transcripts, and diagnostic in ways the standard political commentary does not name.
Eight Prime Ministers, one pattern
The framework does not treat all eight recent PMs equally, because their patterns are not equally rich. Some are diagnostically noisy or short-tenured to the point of being data-thin. Others are textbook cases of the framework's structural argument. The piece treats each by the merits of the pattern rather than by tenure length or political prominence.
Tony Blair, from 1997 through roughly 2003, ran one of the most we-coded speech patterns of any post-war PM. His early conference speeches, his 1997 election-night address, his early cabinet appearances read as institutional voice — this country, our people, we can. The framework's diagnostic reading of that period matches what the polling and internal party dynamics recorded: high trust, coherent authority, government functioning. From roughly the Iraq War onward, Blair's register shifted measurably. The I have concluded, I have decided, I take responsibility constructions rose in his speeches through 2004–2007. The pronoun shift preceded the polling collapse by approximately eighteen months. The framework would have flagged the authority-erosion earlier than the standard commentary did.
Gordon Brown is the framework's cleanest example of a leader whose pronoun pattern was structurally unfavourable from the beginning. His speeches, going back to the Treasury years, ran heavily first-person-singular even when he was announcing collective initiatives. I have decided, I believe, my judgement. This was not merely a matter of personal style; it correlated with a premiership that struggled with authority, coherence and the ability to project institutional confidence. The framework would not have predicted Brown's specific failures, but it would have flagged the structural fragility his pronoun pattern implied. The academic literature is clear that heavy-I patterns are less an indicator of confidence than of what Pennebaker calls self-focused attention under threat — a register that emerges when a leader is defending rather than leading.
David Cameron's pattern is a mixed and interesting case. During the coalition years 2010–2015, his speeches were heavily we-coded — the coalition itself forced the institutional-voice register. Post-2015, when the majority government began, and particularly during the Brexit referendum campaign, the register shifted toward I will, I have set out, my view. His resignation speech in 2016 was heavier on the first-person singular than any speech of his coalition years. The framework flagged the trajectory. The Brexit vote confirmed it.
Theresa May is the textbook case. Her speeches from the 2016 leadership contest onward built the entire premiership around the first-person singular. Strong and stable leadership, I will negotiate, I have set out, I will not, I believe. The strong and stable phrase itself — repeated so frequently it became a caricature — was I-adjacent structurally, positioning May as the singular figure to whose judgement the country should defer. The framework's diagnostic reading of May in 2017–2019 is unambiguous: heavy defensive-I, poor institutional coherence, authority eroding in real time. The polling caught up eventually. The pronoun pattern was ahead of it by months.
Boris Johnson is the framework's most interesting hybrid case, and the piece would spend more time on him than on some of the shorter-tenured PMs because his pattern reveals something the simpler cases do not. His speeches were heavily we-coded at the surface — we can, we will get Brexit done, we are the party of the people. But the underneath was heavily I-operational — his decision-making, his political-relationship management, his crisis behaviour all read as first-person-singular in ways his rhetorical register concealed. The framework's diagnostic on Johnson would have been a warning: the pronoun pattern of the speeches did not match the pronoun pattern of the operations, and mismatch itself is a signal the literature identifies as unstable leadership. His trajectory bore this out.
Liz Truss's 49 days do not provide a long enough corpus to draw structural conclusions from, but the pattern was heavy first-person-singular from the leadership contest through the mini-budget: my plan, I will, my growth strategy. The framework does not need much data to flag heavy-I in an early premiership as structurally unfavourable, and Truss's outcome confirmed the prediction the framework would have made from her leadership-contest speeches alone.
Rishi Sunak's pattern was more technocratic-institutional than either heavy-I or heavy-we. His speeches through 2022–2024 ran to the government has, the country needs, our approach — a managerial register that neither performed the confident we of Blair's early years nor collapsed into May's defensive I. The framework's diagnostic on Sunak is that his pronoun pattern was neutral rather than favourable — not a leader whose authority was eroding in the pronoun tell, but not a leader whose authority was compounding either. The polling largely bore this out: consistent but uninspiring, competent-seeming but distant.
Starmer's trajectory, in his own pronouns
Keir Starmer's early premiership through late 2024 carried the institutional-we register. Cabinet appearances, conference addresses, the first six months of Commons statements read as we will, our government, this country deserves. The framework's diagnostic reading of that period was favourable — the pronoun pattern of a leader consolidating authority through collective framing.
Through 2025 and into 2026, the register shifted measurably. The Commons statements on the winter fuel decision, the welfare reforms, the various U-turns that punctuated the second year of the government, all carried a rising proportion of first-person-singular constructions. I take responsibility, I have decided, I believe, my government. The shift was not sudden; it was gradual and observable across sequential speeches. It preceded the polling collapse of spring 2026 by several months, which is the pattern the framework's academic foundation predicts.
The resignation speech itself, delivered on 22 June 2026, is the pattern's endpoint. In prepared remarks running about ninety seconds, Starmer used the first-person singular approximately nine times, first-person plural once. This is not a defect of the speech; it is the register the framework expects at the moment of authority relinquishment. The speech is heavy-I because the speaker is, at that moment, defending a set of decisions in his name that the collective he was leading has withdrawn its support from. The pronoun register and the political reality are aligned. The framework did not need the resignation to predict where the trajectory was pointing. The trajectory had been visible in the speech register since mid-2025.
Burnham's Mayor register, and the Westminster question
Andy Burnham's pronoun pattern, drawn from his eight-year tenure as Mayor of Greater Manchester and from his two major speeches since re-entering Westminster — the Makerfield victory speech of 19 June 2026 and his 'Westminster is broken' policy speech of 28 June 2026 — sits at the far collective-leaner end of the spectrum. "Everyone knows that politics is not working. Everyone can feel that the country isn't where it should be." The Makerfield victory speech opened on second-person collective and stayed there. His policy speech built the entire pitch around collective and place-based framing — we, our regions, this country, every British postcode, place first, Manchesterism. The No. 10 North concept is itself a collective-institutional device rather than a personal one; it distributes authority geographically rather than concentrating it in the person of the PM. The framework's diagnostic reading of Burnham's pattern in June 2026 is that his register is more heavily we-coded than any recent Labour leader at the equivalent career stage, and more we-coded than most Conservative leaders as well.
This is where the framework earns its predictive value. The prediction is bounded and testable. If Burnham's pronoun pattern in his first six months of premiership continues at the collective-leaner end of the spectrum — if his PMQs responses, his Commons statements, his cabinet-level addresses maintain the we, our, this country, our regions register that has characterised his Mayor tenure and his June 2026 leadership pitch — the framework predicts a different trust trajectory than any PM since 1997 has produced. The authority-erosion pattern that every recent PM has eventually settled into would, in this scenario, be delayed or attenuated, because the pronoun architecture that has historically preceded it would not be in place.
If, on the other hand, his pattern shifts to defensive-I within the first six months — as it has for every recent PM whose authority came under pressure — the framework predicts the same trajectory of trust erosion that Blair, Brown, Cameron, May, Johnson, Truss and Starmer each eventually produced. The first fifty PMQs of any Burnham premiership would provide the data. The pattern would be visible in the transcripts before the polling caught up.
The framework does not predict whether Burnham will be a successful PM in the ways political commentary usually measures — policy achievement, electoral outcome, historical reputation. It predicts something narrower and more mechanical: whether the pronoun pattern that has preceded authority erosion in every recent PM will emerge in Burnham's speech in the same window, and whether the trust-erosion pattern that has followed that pronoun shift will follow again. Both are testable. Both will be visible in real time. The framework's contribution is naming what to watch for.
What the pattern is, and what it isn't
The framework's academic foundation supports the following claims. Pronoun patterns are stable across a leader's career, more so than most other verbal markers. Shifts in the ratio are meaningful, particularly under stress. Heavy-I patterns correlate with self-focused attention, defensive posture and organisational fragility. Heavy-we patterns correlate with institutional confidence, distributed authority and trust compounding. These correlations have been observed across CEOs, political leaders, therapy patients and populations under crisis. They are neither deterministic nor the only variable that operates, but they are consistent enough that a leader whose pronoun pattern is shifting toward the defensive-I end of the spectrum is producing a leading signal that the public trust readings will eventually catch up to.
What the framework does not do is predict specific outcomes, adjudicate the moral character of leaders, or serve as a partisan diagnostic. Heavy-I patterns are not a character failure; they are a structural signal. May's heavy-I was not a moral defect but a register that emerged from the specific pressures of her premiership. Blair's heavy-we in 1997 was not virtue; it was the register a consolidating leader produces. Burnham's current we-heavy register is not evidence of superior leadership in advance; it is a pronoun pattern that has, in his case, been more collective-leaning than most Westminster leaders' patterns are, and the framework predicts specific things about what happens if that pattern holds or shifts under the pressures of premiership.
The same observation appears in the framework's other applications — the I-We Index as applied to S&P 500 CEOs produced a similar structural claim about the temperament that runs institutions. The maker of a leader's speech is producing the master. The pattern the pronouns produce is the meaning. The framework's contribution is treating what almost every reader hears as background register as, instead, a mechanically observable leading indicator of leadership trajectory. Applied to eight British PMs, it holds. Applied to Burnham, it names the question rather than answering it. The first six months of his premiership, if it arrives, will produce the data the framework needs to make its prediction visible to the standard commentary that has not yet learned to read the tell.