Read enough apology emails and you start to notice the tell. The sentences get shorter. The vocabulary contracts. A writer who normally reaches for precise verbs suddenly leans on “issue,” “situation,” “things.” Somewhere in the second paragraph the word “I” either vanishes or floods the page, and the difference between those two patterns says more about the person’s state than anything they actually claim. Editors who work on this kind of correspondence learn to read the pressure before they read the content.
This is the strange part of working closely with other people’s prose: the most revealing words are almost never the ones the writer is thinking about. Nobody chooses their pronouns deliberately. Nobody decides, mid-crisis, to drop their use of articles or to switch from the present tense to the past. These choices happen below the level of intention, which is exactly why they carry information. Under pressure, the controlled layer of writing thins out and the automatic layer shows through.
The words you don’t notice are the ones doing the work
The psychologist James Pennebaker spent decades arguing that the small, structural words of a language — pronouns, prepositions, articles, conjunctions — reveal more about a person than the content words everyone pays attention to. In a 2010 review with Yla Tausczik, “The Psychological Meaning of Words,” he and his co-author lay out the case: these function words are processed almost entirely unconsciously, they make up a tiny fraction of our vocabulary but a huge share of the words we actually use, and they track attention, status, and emotional state with measurable reliability.
The mechanism matters here. Content words — negotiation, deadline, refund — are what a writer consciously selects to convey meaning. Function words are the connective tissue that the brain assembles on the fly. You can deliberate over whether to call something a “delay” or a “setback.” You cannot deliberate, in real time, over how often you say “I.” That is why the function words leak. They are too fast and too frequent to manage, and pressure speeds everything up.
Stress flattens the sentence before it touches the argument
The first thing pressure does to writing is narrow it. When the cognitive system is loaded — by anxiety, by time, by the sense of being judged — the linguistic range available to a writer shrinks. This is not a metaphor. In a 2014 study in Psychophysiology, “Speaking under pressure,” Laura Saslow and colleagues found that people whose bodies reacted more strongly to a stressful speaking task — higher heart rate, higher cortisol, more reported distress — produced speech with measurably lower linguistic complexity. The more reactive the speaker, the simpler the sentences.
Editors see the written version of this constantly. A draft written calmly and a draft written against a wall, by the same person, do not read like the work of one writer. The pressured draft has shorter clauses, fewer subordinate constructions, a thinner spread of vocabulary, more repetition of the same handful of words. The argument may be identical; the surface has collapsed. It is tempting to read this as carelessness, but it is closer to triage — the writing mind, short on working memory, spends what it has on getting the point out and lets the structure go.
This is worth separating from quality. Simpler prose is not worse prose, and some of the best editing strips complexity out on purpose. But complexity that disappears involuntarily is a signal, not a style, and a reader picks up on it even without naming it. The flattened sentence reads as strained because it is.
Pronouns mark where the writer’s attention has gone
The most studied of these signals is the first-person singular. The common assumption — that confident, powerful people talk about themselves more — is backwards. Pennebaker’s work on email and conversation, summarised in his book The Secret Life of Pronouns and described in a profile in Yale Scientific Magazine, found that the person of higher status in an exchange uses “I” less. The lower-status party uses it more. His explanation is about attention: the person in the secure position is focused outward, on the task and on the other people; the person who feels exposed is focused on themselves, and the pronouns follow the focus.
Under pressure this becomes legible in correspondence. Watch what happens to “I” in a tense thread. A writer who feels on the back foot produces a wall of self-reference — “I just want to say,” “I think,” “I feel like,” “I’m not sure if I.” A writer who feels in command barely uses the word; their attention is on the problem and the reader. Neither writer chose this. It is the pronoun reporting on where the writer’s mind actually is, regardless of the posture the sentences are trying to project.
This is also why advice to “sound more confident” by editing pronouns mostly fails. You can delete the “I”s, but you are editing the symptom. The reader is responding to a whole cluster of cues — sentence rhythm, hedging, tense, how much the writing attends to them versus to the writer — and a cleaned-up pronoun count over an anxious draft is unlikely to change how the whole thing reads.
The harder a story is to hold together, the more the words give way
There is a related pattern in writing that is doing something effortful in a different sense: constructing an account that the writer does not fully stand behind. In a 2003 study, “Lying Words,” Matthew Newman, Pennebaker and colleagues compared truthful and deliberately false statements and found a consistent linguistic profile. Compared with truth-tellers, people producing false accounts used fewer first-person singular pronouns, fewer references to other people, more negative-emotion words, and showed lower overall cognitive complexity. A text-analysis program using these features classified statements correctly around two-thirds of the time when the topic was held constant.
The honest caveat: two-thirds is well above chance and nowhere near a lie detector. This is a statistical tendency across many texts, not a verdict on any single sentence, and the markers overlap heavily with ordinary stress — which is precisely the point worth keeping. The same fingerprint shows up whether someone is fabricating or simply anxious, because both situations load the system the same way. Holding a story together that isn’t true is cognitively expensive, and expense shows up as the same flattening that pressure produces. Distance from the self, fewer other-references, a darker emotional tint: these are the marks of a mind working harder than the words admit.
What the reader does with all this
None of these signals require a reader to know the research. People absorb them automatically, the same way the writer produced them. A hiring manager reads a cover letter and finds it “trying too hard” without being able to say why. A colleague reads a Slack message and senses defensiveness in something that, line by line, is perfectly polite. The judgment is real and usually quick, and it is built from exactly the structural features the writer never consciously placed: the density of “I,” the hedges, the tense, the contraction of range.
Tausczik and Pennebaker make a further point that complicates any tidy reading of this. Function words are also how people coordinate — two writers who are in sync tend to converge in their use of these small words, a phenomenon they call linguistic style matching. So the words under pressure don’t only broadcast a state; they shape the relationship between writer and reader in real time, pulling the exchange tighter or pushing it apart.
This is the part that should make any writer slightly uneasy, and any editor slightly more careful. You can revise the content all you want; the structure underneath is reporting on you the whole time, and the reader, without ever deciding to, is listening to it.
