Rules
Part of Why humorous quotes lose their authors
Humorous quotes source check: building the search query
A humorous quotes source check starts by splitting a joke into frame, pivot and wording, then querying the words that survive retelling.
A joke cannot be searched like a quotation. Its wording changes with every teller, so the sentence you typed into the search box may never have been printed by anyone. A humorous quotes source check therefore builds its query from the parts of a joke that do not change.
That means searching a structure, a setup shape and a turn, in corpora that date their contents. Quotation collections are rarely the right place to look.
What to take away
- Split the joke into frame, pivot and wording before you type anything.
- Query the odd noun, the number or the incidental proper noun. Leave out the punchline and the famous name.
- Run the same query in dated corpora, with newspaper filler first.
- Old newspaper text is machine-read and left uncorrected, so a miss is not proof of absence.
- Read a single early hit as a ceiling on the date, and a memoir as a recollection.
Split the joke before you search
Break the joke into three parts before you search anything: the frame, the pivot and the wording. Only the first two are stable enough to query.
Break the joke into three parts
- Framesituation, who where doing what
- Pivotthe turn, reversal, misunderstanding
- Wordingunstable, assume every word changes
- Query frame and pivot in plain words
- Query wording finds copies of your page
The frame. The situation: who is where, doing what. This is the most stable element, and it is often what makes the joke recognizable across two centuries and four languages.
The pivot. The thing that turns: the misunderstanding, the reversal, the literal reading of a figure of speech, the escalation. Also stable, because it is the joke.
The wording. Unstable. Assume every word can change and that at least half of them have.
Queries built from the frame and the pivot, using the plainest possible words, will find relatives of your joke. Queries built from your wording will find copies of the page you got it from.
Pick the words that survive
Some elements resist rewriting because the joke stops working without them.
- The concrete object.A joke that turns on a particular item keeps that item. Retellers change adjectives, not the thing on the table.
- The occupation or role.Frequently stable within a run of retellings, though it migrates across generations, which is itself informative.
- The number.Jokes with a count in them keep the count.
- The odd preposition or word order.Nobody remembers editing the connective tissue, so it survives even as the vivid words rotate.
- The proper noun that is not the famous name.A place, a title, a product category. Retellers replace the celebrity in the setup far more readily than the incidental noun.
Combine two of these in one query, without the punchline and without the famous name, and you have the best chance available of reaching an early appearance.
That is the same principle as the dull anchor in ordinary source checking, pushed further because there is even less stable text to work with.
Build the query in five steps
- Write the joke in the plainest words you can, as you first heard it.
- Mark the frame and the pivot.
- Pick two surviving elements, such as the object and the number.
- Delete the punchline and the famous name from the query.
- Run it in a dated corpus and note the corpus and the years searched.
If the first pair returns only copies of the page you started from, swap one element for another and run it again. Wordplay is the one exception: the words the pun needs are load-bearing, so keep both of them.
Run the query where jokes were printed
Search corpora that date their contents. Newspaper filler is the richest seam for antedating a comic line: for a long stretch of print history, short jokes filled column space, were swapped between papers and ran without credit.
For U.S. papers, start with Chronicling America, a searchable digital collection of historic newspaper pages through 1963 from the Library of Congress. The newspaper filler it holds is where many jokes first show up.
Productive corpora vs quotation collections
Productive corpora
- Dating
- Usually dated
- Credit
- Printed without credit
- Coverage
- Filler, humor papers, almanacs
- Value
- Antedates the line
Quotation collections
- Dating
- Often undated
- Credit
- Assigned to a name
- Coverage
- Compiled for readers
- Value
- Repeats the attribution
| Corpus | What it holds | Why it dates a joke |
|---|---|---|
| Humor and trade papers | Joke columns for performers and general readers | Their dates are usually reliable |
| Almanacs, annuals, after-dinner speech books | Material compiled for people who needed it | Often dated |
| Reviews of performances | A reviewer's paraphrase of a joke | Shows something like it was said, on a date, in a place |
| Scripts and cue material | Rarely public | Decisive when available, but different from what was broadcast |
| Recordings and broadcast listings | Performances of the last century | A hard date, and the delivery as well |
For books and annuals, Google Books and the Internet Archive are searchable starting points. Yale Book of Quotations can offer cited leads, but check those citations against the printed source.
A compact example is Oscar Wilde's line, "I can resist everything except temptation": it appears in Lady Windermere's Fan, published in 1893. That gives a dated printed appearance for the line, not proof that Wilde originated the joke or that no earlier version existed.
Expect gaps in old newspaper text
Searching old newspapers means searching machine-read text, and a missed hit is not proof that a joke was never printed. Pages are scanned and then processed for full-text searching through Optical Character Recognition, and the resulting text is left uncorrected.
So try a second pair of surviving words and a variant spelling before you accept a blank result. When nothing comes back after that, record the archive and the years, because that is all the null result can claim.
Read the result set
| What comes back | What it means | How to say it |
|---|---|---|
| Many undated appearances with no name, then a famous name later | Standard attribution drift | Circulating anonymously in print by a given year; the name appears only afterwards |
| The same joke with different names in different decades | Migration between wits | Credited to several people over time; no early attribution located |
| The same structure with different targets or occupations | A joke type rather than a text | A version of a much older joke, with the target changed |
| One dated performance and nothing earlier | Possibly original, possibly the first surviving trace | Earliest located appearance, with the corpus named |
| Everything traces to one anthology | The anthology is where the attribution begins | The credit originates with that compiler, who cites nothing |
| A hit that predates the credited person entirely | Settled | In print before the person credited was working |
For a joke type there may be no author to find. The honest result is then a description of a family of jokes and roughly when it starts showing up.
Two traps in a single hit
The reprint trap. An archive gives you the date of the issue you are looking at, and that issue may be reprinting a joke that ran elsewhere a fortnight or a decade earlier.
Filler moved constantly. Treat any single early hit as a ceiling on how late the joke can be, not as its birthday, and look for whether the item credits another paper.
Two traps in early evidence
Reprint trap
- Source
- Archive issue date
- Problem
- May reprint earlier item
- Check
- Does it credit another paper?
- Use it as
- Ceiling, not birthday
Late testimony trap
- Source
- Memoir or interview
- Problem
- Recollection, polished
- Check
- Date it to the telling
- Use it as
- Evidence of memory
The late testimony trap. A comedian's memoir, or an interview given years afterwards, is a recollection. Performers polish their own histories in good faith, and the story of when a line was first used tends to improve alongside the line. It is evidence about what the person remembered, dated to the telling.
Write the result as a date
For most jokes the strongest supportable claim is chronological, so write the chronology.
Writing the chronology
- Earliest appearance found, and corpus
- Credit attached, if any
- When the famous name first attaches
- Archives searched and periods covered
- Absence from a database is not absence
That sentence is more useful than a confident credit or a flat denial. It survives the next person checking it, which is the real test.
The Misquotes and Fact Checks resource covers the vocabulary for stating what a check established. The Humorous Quotes overview sets out which credits are available when one person writes material and another performs it.
Common questions
How do I search a joke that depends on wordplay?
Search the words the pun requires, both of them, without the rest of the sentence. Wordplay is the one case where the exact words are load-bearing, so they change less, which makes these jokes easier to trace than most.
The joke exists in another language. Does that help?
Considerably. A structure attested in several languages is old, in the way anything carried by oral tradition is old, and translated joke books are a real corpus. It also usually rules out any modern attribution, which is often the whole question.
Nothing comes back at all. What does that license?
Only that this structure, in the forms you tried, is not in the archives you searched, for the years they cover. Comic material is under-digitized, spoken material barely at all, and a null result here is weaker than a null result for printed prose. Say which archives, and say the years.
