Steven Pinker
Harvard cognitive psychologist, linguist and author of The Sense of Style
Where to hear and read about Steven Pinker1 episode or test
Steven Pinker is a Harvard cognitive psychologist and linguist who has written nine books, including The Sense of Style, and spent his working life studying language and cognition. The thing most worth learning from him is to assume the reader knows less than you do, then test every draft on a smart outsider and squeeze it to a hard word limit.
The curse of knowledge is the difficulty that we all have in knowing what it's like not to know something that we know.
Harvard Professor Explains The Rules of Writing (Steven Pinker Interview)
Steven Pinker: the story and the lessons
1Why Steven Pinker blames bad writing on the curse of knowledge
Pinker says bad writing is usually forgetting what the reader doesn't know. He rejects the theory that academics hide thin ideas in jargon. He cites Hanlon's razor and knows many brilliant people who have plenty to say and no wish to obfuscate. He calls the curse of knowledge the single biggest flaw in writing and communication.
It shows up as acronyms, jargon and abstractions. Egocentrism and the absence of a theory of mind are other names for the same flaw.
A molecular biologist lost a TED audience in about four seconds. He launched into experiments in his usual lab jargon without saying what problem he was solving or why it mattered. Several hundred people were lost, and everyone except him could tell.
Pinker's fix is to open by naming the problem and why it is significant before presenting any findings.
Pinker tests drafts on real readers because imagining them is not enough. He used to show drafts to his mother, who was bright and well read but not a cognitive psychologist. Now he uses his editor and academic friends in other fields, who often say they have no idea what he means.
You cannot tell when the curse hits you. Even in his own department, students hand him thesis proposals he cannot follow, because each lab of five or six people has absorbed its own jargon.
He writes for curious, educated readers who are not his peers. Pinker does not write for a random sample of the population or for specialists in his own field. He writes for people who are intellectually curious and have some education.
Choosing that reader tells him which terms to explain and which to skip.
He defines or replaces the shorthand his specialty uses every day. Terms like ecosystem, species or potentiation are essential inside a field, because two syllables can carry a huge body of knowledge. The trouble starts when a specialist writes for outsiders and forgets the terms are not basic to them.
His answer is to back up and define the term, or use a picture instead.
2How Pinker makes writing visual and concrete
Pinker writes things the reader can picture, not abstract nouns. Understanding is often visual, bodily, emotional or auditory, so language works best when it points at something sensory. He says to write 'bunny rabbit', not 'stimulus', and to avoid level, perspective, framework, paradigm and concept.
A sentence about stimulus intensity and reaction really means kids look longer at a bunny than at a truck. In his words, "no one can form an image of a paradigm in their mind's eye."
He pairs every generalization with a concrete example. A generalization alone is vague and an example alone is pointless. Pinker explains that a word's meaning need not match its parts by listing bathroom, breakfast and Christmas.
In his own field he says he cannot tell what nine out of ten generalizations mean when they come without examples. He says "a generalization erases detail."
Pinker puts on the page the context a conversation would supply. Speakers can say 'she' or 'the thing' because they share common ground and can see a furrowed brow. A reader may live in another country, or be reading after the writer is dead.
This is why speaking comes naturally and writing has to be taught. Writing strips away shared context and live feedback.
He reaches for images from shared experience before technical terms. Older prose feels vivid because writers lacked centuries of academic abstractions and had to use images everyone knew. Where we write 'aggression', they might have evoked a hawk.
Pinker treats this as a habit anyone can borrow by looking in common knowledge before reaching for a coined term.
Pinker treats children's plain explanations as raw material. His grandson said clouds are water vapor and smoke is fire vapor. Children have no mass of jargon or cliches, so they describe what they can see.
3Pinker's rules for brevity and sound
Pinker squeezes pieces into a hard word limit. When a newspaper or magazine gives him a fixed word count, he compresses the piece himself rather than let an editor chop it. Fitting an 800 word limit improved the prose even when he thought the draft was already perfect.
Short limits push him toward concrete language and away from woolly idioms and cliches. He says it improves the prose "as if by magic".
He treats every extra word as work the reader must do. Pinker cites Hamlet's line that brevity is the soul of wit and E. B. White's first rule, omit needless words. Each syllable costs the reader cognitive effort.
The rule contains no needless words itself. White's Cornell professor repeated it when he ran out of other advice.
Pinker reads his prose aloud, or at least mumbles it. If he cannot say a sentence smoothly, he reasons, the reader will not sound it out smoothly in their head either.
He says he did not invent the tip and calls it highly recommended.
He listens for rhythm, trims sibilants and uses light alliteration. Disrupting a sentence's beat interferes with reading as well as speech. He swaps in a synonym when too many s sounds pile up.
A little alliteration helps a sentence roll past more easily, but it must not be conspicuous or it feels forced.
Pinker keeps jokes short and never telegraphs the punchline. Humor depends on brevity and freshness. A strained joke gets groans, and a joke heard 37 times is stale.
Comedians pare jokes down, because dragging one out steps on the punch line or gives it away.
He studies the classics to build an ear for good prose. Edison, Darwin and Oliver Wendell Holmes Sr. were not professional writers yet wrote beautifully. Pinker says they trained on the classics, used prose to present themselves, and had no ready-made cliches.
He quotes Edison on how electricity would liberate women in Enlightenment Now.
4Why Pinker holds academics to a higher standard
Pinker sees bad academic prose as a waste of brilliant, publicly funded work. Smart scholars write for a small closed circle, though taxpayers fund the research. Readers must re-read a paragraph five or six times and may still get it wrong.
He meets this when reviewing grants, peer-reviewing manuscripts and evaluating people for tenure. It gets under his skin when academics pour brain power into scholarship and then blow off telling the world what they did.
He says linguists who write badly ignore their own research. Linguistics and psycholinguistics have many bad writers, though they study what makes a sentence hard to understand. Pinker tries hard to be clear even in his academic writing.
5What Pinker thinks AI does to prose
Pinker expects AI prose to be clear but generic. Model output is plain, orderly and free of jargon, with an intro and a conclusion, which beats most academic or legal prose. But it is so generic you can almost recognize it.
He guesses that human fine-tuning pushes it toward the five-paragraph essay. He also guesses that averaging prose removes the worst constructions, as morphing a yearbook's faces gives an attractive face. He calls this completely speculative.
He would now give pattern extraction more weight in explaining language. His training in computational cognitive science, Chomskyan linguistics and classic AI centered on rules, algorithms and logic. He found it hard to imagine a neural network drawing sensible prose from a huge unstructured mass of text.
He does not conclude the mind is a language model. A child would need about 30,000 years of listening to learn as a model does, and children also interact with the world and work out what speakers mean.
Modern prose turned conversational long before AI arrived. Over a century, dress and address loosened, and authenticity and spontaneity became the ideal. Fancy language now reads as pompous, stilted and distancing.
Pinker got detention in third grade for calling a teacher by her first name.
Harvard Professor Explains The Rules of Writing (Steven Pinker Interview)
Pinker argues that most bad writing comes from the curse of knowledge, forgetting that readers do not know what you know, and that the fixes are concrete images, examples, brevity and testing drafts on real people. The most useful habit is to show every draft to a smart reader outside your field and to squeeze it into a hard word limit.
Pinker says bad writing is usually forgetting what the reader doesn't know. He rejects the theory that academics hide thin ideas in jargon. He cites Hanlon's razor and knows many brilliant people who have plenty to say and no wish to obfuscate. He calls the curse of knowledge the single biggest flaw in writing and communication.
It shows up as acronyms, jargon and abstractions. Egocentrism and the absence of a theory of mind are other names for the same flaw.
A molecular biologist lost a TED audience in about four seconds. He launched into experiments in his usual lab jargon without saying what problem he was solving or why it mattered. Several hundred people were lost, and everyone except him could tell.
Pinker's fix is to open by naming the problem and why it is significant before presenting any findings.
Pinker tests drafts on real readers because imagining them is not enough. He used to show drafts to his mother, who was bright and well read but not a cognitive psychologist. Now he uses his editor and academic friends in other fields, who often say they have no idea what he means.
You cannot tell when the curse hits you. Even in his own department, students hand him thesis proposals he cannot follow, because each lab of five or six people has absorbed its own jargon.
He writes for curious, educated readers who are not his peers. Pinker does not write for a random sample of the population or for specialists in his own field. He writes for people who are intellectually curious and have some education.
Choosing that reader tells him which terms to explain and which to skip.
He defines or replaces the shorthand his specialty uses every day. Terms like ecosystem, species or potentiation are essential inside a field, because two syllables can carry a huge body of knowledge. The trouble starts when a specialist writes for outsiders and forgets the terms are not basic to them.
His answer is to back up and define the term, or use a picture instead.
Pinker writes things the reader can picture, not abstract nouns. Understanding is often visual, bodily, emotional or auditory, so language works best when it points at something sensory. He says to write 'bunny rabbit', not 'stimulus', and to avoid level, perspective, framework, paradigm and concept.
A sentence about stimulus intensity and reaction really means kids look longer at a bunny than at a truck. In his words, "no one can form an image of a paradigm in their mind's eye."
He pairs every generalization with a concrete example. A generalization alone is vague and an example alone is pointless. Pinker explains that a word's meaning need not match its parts by listing bathroom, breakfast and Christmas.
In his own field he says he cannot tell what nine out of ten generalizations mean when they come without examples. He says "a generalization erases detail."
Pinker puts on the page the context a conversation would supply. Speakers can say 'she' or 'the thing' because they share common ground and can see a furrowed brow. A reader may live in another country, or be reading after the writer is dead.
This is why speaking comes naturally and writing has to be taught. Writing strips away shared context and live feedback.
He reaches for images from shared experience before technical terms. Older prose feels vivid because writers lacked centuries of academic abstractions and had to use images everyone knew. Where we write 'aggression', they might have evoked a hawk.
Pinker treats this as a habit anyone can borrow by looking in common knowledge before reaching for a coined term.
Pinker treats children's plain explanations as raw material. His grandson said clouds are water vapor and smoke is fire vapor. Children have no mass of jargon or cliches, so they describe what they can see.
Pinker squeezes pieces into a hard word limit. When a newspaper or magazine gives him a fixed word count, he compresses the piece himself rather than let an editor chop it. Fitting an 800 word limit improved the prose even when he thought the draft was already perfect.
Short limits push him toward concrete language and away from woolly idioms and cliches. He says it improves the prose "as if by magic".
He treats every extra word as work the reader must do. Pinker cites Hamlet's line that brevity is the soul of wit and E. B. White's first rule, omit needless words. Each syllable costs the reader cognitive effort.
The rule contains no needless words itself. White's Cornell professor repeated it when he ran out of other advice.
Pinker reads his prose aloud, or at least mumbles it. If he cannot say a sentence smoothly, he reasons, the reader will not sound it out smoothly in their head either.
He says he did not invent the tip and calls it highly recommended.
He listens for rhythm, trims sibilants and uses light alliteration. Disrupting a sentence's beat interferes with reading as well as speech. He swaps in a synonym when too many s sounds pile up.
A little alliteration helps a sentence roll past more easily, but it must not be conspicuous or it feels forced.
Pinker keeps jokes short and never telegraphs the punchline. Humor depends on brevity and freshness. A strained joke gets groans, and a joke heard 37 times is stale.
Comedians pare jokes down, because dragging one out steps on the punch line or gives it away.
He studies the classics to build an ear for good prose. Edison, Darwin and Oliver Wendell Holmes Sr. were not professional writers yet wrote beautifully. Pinker says they trained on the classics, used prose to present themselves, and had no ready-made cliches.
He quotes Edison on how electricity would liberate women in Enlightenment Now.
Pinker sees bad academic prose as a waste of brilliant, publicly funded work. Smart scholars write for a small closed circle, though taxpayers fund the research. Readers must re-read a paragraph five or six times and may still get it wrong.
He meets this when reviewing grants, peer-reviewing manuscripts and evaluating people for tenure. It gets under his skin when academics pour brain power into scholarship and then blow off telling the world what they did.
He says linguists who write badly ignore their own research. Linguistics and psycholinguistics have many bad writers, though they study what makes a sentence hard to understand. Pinker tries hard to be clear even in his academic writing.
Pinker expects AI prose to be clear but generic. Model output is plain, orderly and free of jargon, with an intro and a conclusion, which beats most academic or legal prose. But it is so generic you can almost recognize it.
He guesses that human fine-tuning pushes it toward the five-paragraph essay. He also guesses that averaging prose removes the worst constructions, as morphing a yearbook's faces gives an attractive face. He calls this completely speculative.
He would now give pattern extraction more weight in explaining language. His training in computational cognitive science, Chomskyan linguistics and classic AI centered on rules, algorithms and logic. He found it hard to imagine a neural network drawing sensible prose from a huge unstructured mass of text.
He does not conclude the mind is a language model. A child would need about 30,000 years of listening to learn as a model does, and children also interact with the world and work out what speakers mean.
Modern prose turned conversational long before AI arrived. Over a century, dress and address loosened, and authenticity and spontaneity became the ideal. Fancy language now reads as pompous, stilted and distancing.
Pinker got detention in third grade for calling a teacher by her first name.