Notes
Stylistic notes
Two findings about prose style, with the caveats up front.
The project’s main story is in the catalogue: who quoted Shakespeare, how often, and from which plays. Two other findings emerged from finer-grained statistical analysis. Both are real, both are measurable, but both require more methodological context than a direct quotation does, and each carries an honest caveat about what it does and doesn’t show. They live here, on a separate page, so they don’t crowd out the more direct evidence on the homepage and the catalogue.
If the catalogue is the project’s testimony, the two findings below are circumstantial. Read them as that.
Finding 1
Franklin’s prose is late-Stuart English, not Shakespeare in particular
An earlier version of this project framed Benjamin Franklin as “the Founder who sounds most like Shakespeare without ever quoting him.” That framing is too strong. What the data actually shows is that Franklin’s prose retains seventeenth-century English features at higher rates than any of the other five Founders — archaic verb inflections, the contraction ’tis, the discourse marker methinks, the second-person thou outside of biblical quotation. Shakespeare also retains those features (he’s writing in early-modern English). The two prose registers overlap, but the overlap is with late-Stuart English generally, not with Shakespeare specifically.
The honest framing: Franklin (born 1706) learned to write at sixteen using older models. His prose register was set by seventeenth-century English, the same English Shakespeare wrote in a century earlier. That’s a generational fact about Franklin, not a finding about Shakespearean influence on him.
The measurable signal is real. Franklin uses ’tis at 390 occurrences per million words. The next-highest Founder is around 154/M. Shakespeare uses it at roughly 1,608/M. Franklin’s rate is roughly two and a half times every other Founder’s and about a quarter of Shakespeare’s. He shows the same pattern on hath, on the second-person archaic forms, and on the older modal verbs.
What the data can’t show is the counterfactual: would Franklin’s prose look like this if Shakespeare had never existed? Almost certainly yes — the seventeenth-century English Franklin absorbed at sixteen wasn’t Shakespeare’s alone; it was the working register of the periodicals and pamphlets he was reading. So the finding is best stated as a feature of generation and reading material, not of Shakespearean inheritance.
That said, the contrast with Adams (born 1735, 29 years younger) is still informative. Adams’s archaic-form rate is significantly lower than Franklin’s, even though they overlap in writing years (1750–1820 for Franklin, 1750–1826 for Adams). So there’s a Franklin-specific effect on top of the generational one. The case study below works through this in more detail.
Finding 2
The collocational worlds around shared abstract nouns diverge
This finding is often misread on first encounter, so the framing matters: the project is not claiming that Shakespeare owned the words honour, power, love, friend, virtue, death. Both corpora use those words constantly. The Founders use honour roughly 18,600 times; Shakespeare uses it 726 times. Same word, both writers, no monopoly anywhere.
What the analysis actually measures: For each shared abstract noun, the project extracts the words that habitually appear within five words of the target in each corpus — the noun’s collocational neighbours. The neighbours diverge sharply. Honour in Shakespeare collocates with pawn, mine, lord, Brutus: the vocabulary of staking and challenging. Honour in the Founders collocates with sir, esteem, letter, excellency: the vocabulary of letter-closing protocol. Same noun, two completely different conceptual worlds.
The project frames this as a finding about conceptual inheritance: the Founders inherited Shakespeare’s lexis but rebuilt the conceptual content of words like honour, power, love, virtue. They use the same English words; the words mean appreciably different things in their writing than in his.
The pattern holds for fourteen abstract nouns tested. The interactive explorer lets you flip through them and see the contrasts directly. The collocate columns are ranked by Stefanowitsch’s G log-likelihood measure with Bonferroni multiple-comparisons correction; full statistical detail is in the paper.
The legitimate skepticism about this kind of finding is that the contrast isn’t between Shakespeare and the Founders so much as between drama and political correspondence. The letter-closing-protocol use of honour is a feature of eighteenth-century epistolary convention, not of Founder thought specifically; an English merchant’s letters of the same period would probably show the same pattern. So the collocational divergence is a genre signal at least as much as a conceptual-inheritance signal.
That said, the divergence is real and runs across fourteen target nouns. The case study below works through honour in detail.
Where these fit
The relationship to the rest of the project
Both findings live in this notes section rather than in the main case-study sequence because they’re the project’s most easily over-claimed findings. The catalogue’s direct quotations and by-name references stand or fall on whether the passage in question actually contains the Shakespearean text. The two findings here stand or fall on whether you accept the underlying statistical comparison — archaic-form rates in the case of Franklin, differential-collocate distributions in the case of honour. They require trusting more of the methodology to take the claim at face value.
They’re here for readers who want to know what the corpus shows about prose register and conceptual vocabulary, with the caveats stated honestly. The main argument of the project — that Adams quoted Shakespeare prolifically across six decades and that Hamilton wrote 2.2 million words without naming him once — doesn’t depend on either of these findings.
Back to the homepage; read the full Two Modes essay for the broader argument; browse the full catalogue.