Style: Lessons in Clarity and Grace
Joseph Williams · 1981 · Copyrighted — cite only, no excerpts
Williams explains why sentences feel clear or muddy, and gives you the structural analysis to fix them. Two principles do most of the work. First: put old information before new (the given-new contract). Second: use verbs, not nominalizations. “The investigation of the problem” is muddy. “We investigated the problem” is clear. Information flows from known to unknown. Violate the flow and the reader stalls.
The argument
Joseph Williams was a linguist at the University of Chicago. His book, first published in 1981 as Style: Ten Lessons in Clarity and Grace, is the most analytical of the major style guides. Williams gives mechanisms. He explains what clarity is, structurally, and why certain sentences feel muddy while others don’t.
The core insight is that readers process sentences by tracking information flow. A sentence feels clear when it starts with what the reader already knows (given information) and ends with what's new. Violate that order and the reader has to hold unfamiliar material in working memory while searching for the connection. Williams calls this the given-new contract. His contribution was translating functional linguistics research into a writing tool.
The contract also explains why passive voice sometimes works: when the passive puts familiar information in subject position, it satisfies the contract better than the active would.
The second major insight is about nominalizations, verbs turned into abstract nouns. “The establishment of the program” buries the action (“establish”) inside a noun and forces the reader to unpack it. Williams shows that bureaucratic, academic, and legal prose share the same structural disease: nominalizations replacing verbs, abstractions replacing actors. The fix is mechanical. Find the nominalization. Extract the verb. Make the real actor the subject. The sentence clears up.
The core principles
| Principle | The mechanism | What it fixes |
|---|---|---|
| Given-new contract | Put what the reader already knows at the start of the sentence; put new information at the end | Sentences that feel randomly ordered |
| Nominalizations | Convert abstract nouns back to verbs | “The establishment of...” → “We established...”; bureaucratic opacity |
| Characters as subjects | Make the real actors the grammatical subjects | “It was found that...” → “The researchers found...”; missing agency |
| Sentence-level cohesion | End of sentence N connects to start of sentence N+1 | Paragraphs where each sentence feels disconnected |
| Stress position | The end of a sentence carries the most emphasis | Important information buried in the middle |
Revisions
Given-new contract
Before: “A sharp increase in server response times was caused by the new caching layer. The engineering team had deployed it without load testing.”
After: “The engineering team deployed a new caching layer without load testing. Response times spiked.”
The first version leads with “a sharp increase” — new information the reader has no context for. The revision starts with the actor and ends with the consequence. Each sentence begins with what the previous one established.
Nominalization → verb
Before: “The implementation of the new policy resulted in the reduction of customer complaints.”
After: “We implemented the new policy. Customer complaints dropped.”
Two nominalizations (“implementation,” “reduction”) buried the actions. Extract the verbs, make the actors subjects, and the sentence halves in length.
Characters as subjects
Before: “It was determined by the review board that the application had sufficient merit for approval.”
After: “The review board approved the application.”
The original hides the actor (“review board”) behind “It was determined” and buries the action inside “sufficient merit for approval.” One sentence instead of one paragraph.
Discussion
The given-new contract turns voice choice from a rule into a derivation. “Use the active voice” holds when the actor is given information. But when the object is given and the actor is new, the passive puts familiar material first and satisfies the contract. Williams doesn’t override the rule; he explains the conditions under which it applies and when it doesn’t.
The given-new contract also explains why some academic prose is genuinely hard to read and not just jargon-heavy. A paper full of unfamiliar terms isn’t necessarily opaque. If each new term is introduced at the end of a sentence that begins with something the reader already knows, the information lands. The problem is when every sentence leads with new information and buries the connection to prior context.
The contract scales beyond individual sentences. A paragraph coheres when each sentence begins with what the previous sentence made familiar. If a passage feels disjointed, check whether the information flow is broken. The fix is usually reordering, not rewriting from scratch.
The given-new contract draws on functional linguistics: Mathesius and the Prague School described it in the 1930s, Halliday formalized theme and rheme in the 1960s, and Clark & Haviland tested it experimentally in 1977.
Vibelogging — clear information flow is parseable by humans and machines- back to prose writing