How to think about inspirational quotes source check

A source check is not a search for the quotation. It is a search for the oldest appearance of the quotation, and everything about how you run it follows from that one distinction.

Ordinary searching is optimized for the opposite thing. It surfaces what is popular, what is recent, and what matches loosely: all three of which are noise here, because popularity, recency, and looseness are the mechanisms that damaged the line in the first place. This page is about building queries that see past them, and about reading what comes back.

What to take away

Fix the claim before you search

Write down two things and keep them separate.

The string. The exact wording you are testing, punctuation included, plus every variant you have already encountered. You are checking a specific sentence, not a general sentiment, and you will lose track of which version you are chasing within about ten minutes if you do not write them down.

The claim. Who is said to have produced it, and, if anyone says, where and when. Most attributions supply only a name. That absence is your first data point, and our account of the two claims a quotation makes explains why the name and the wording get tested separately.

Building the query

Use the dull anchor. Split the line into the memorable part and the connective part. Retellers rewrite the memorable part; they leave the connective tissue alone, because nobody remembers editing a preposition. Search a distinctive stretch of the boring half in exact-phrase quotation marks.

Drop the punctuation. Em dashes, curly quotation marks, ellipses, and accented characters break exact matching in many archives, and automatic transcription of old print does not render them consistently. Search a run of plain words.

Search each variant separately and compare dates. If one wording consistently returns older results than another, that is a direct observation about which came first.

Search the longer form you suspect. Where the line looks compressed, guess at the missing clause and search for the fragment either side of it. This is how a "quotation" is often revealed as somebody's boiled-down version of a paragraph.

Search the idea in different words. Where the phrasing has drifted completely, only a topical search will find the ancestor.

Search the name and the line together, and separately. The most informative result is often the set of early appearances that carry the line without the name.

Choosing where to look

Archives differ in what they cover and in what they cannot see. Knowing the blind spot is as important as knowing the holdings.

Library digitization projects are the practical entry point for the first two rows, and the Library of Congress research guide to its newspaper collection shows how such a project describes its own coverage. The honest form of any finding names the corpus you used.

Concrete starting points include HathiTrust and the Internet Archive for scanned books, and Chronicling America for digitized United States newspapers. Google Books casts the widest and least reliable net, and Canadiana covers Canadian print. Each publishes a coverage statement, and that statement is what you quote when you report what you searched.

Reading the result set

Sort ascending by date. Then look at the shape of what comes back: the pattern is usually more informative than any individual hit.

The second row is the most common outcome for a widely circulated inspirational line, and it is close to decisive when the gap is large. A quotation that existed for decades without the famous name attached was not produced by the person now credited.

A short worked example. The line below is invented for this walkthrough, and so are the results, so nothing here reports a real attribution.

The string: "A tidy desk is a quiet argument." The claim: a twentieth-century American novelist, no title given. Two variants circulate, "a quiet argument" and "a silent argument".

The dull anchor is "quiet argument", so the first query is that exact phrase in one newspaper archive and one book archive, sorted ascending by date. The variant goes into the same two archives as a separate query.

The earliest hits are mid-century quotation dictionaries, and the novelist's name appears only in collections printed decades later. That is the attribution-drift pattern from the table above, so the verdict is unverified: the wording is documented in reference works, and no primary source carries it.

What a null result licenses

Be precise here, because this is where careful work gets overstated.

Finding nothing means: this wording does not appear in the corpora I searched, for the period they cover, at the accuracy of their transcription, in the forms I tried. That is all.

It does not mean the line was invented. Archives are incomplete, transcription of old print is imperfect, spoken material was rarely recorded, and much of what exists in print was never digitized. A confident announcement that a phrase "does not appear before" a year is only as strong as the corpus behind it.

What a null result does license is a change of label. You can stop presenting the line as a sourced quotation and start presenting it as unverified, which is both accurate and more informative than what you had before.

Write the finding down

Four lines, and they save you and everyone after you from repeating the work:

Cite the archive, not the search page. A later reader needs the archive name, the coverage the project states, and the day you ran the query, because that is what lets them rerun the same search in the same place. A finding that says only that the line was searched online has to be redone from the start.

Then put it somewhere you will find it. The single most wasted effort in this whole activity is doing a good search, reaching a clear result, and having no record of it eight months later when the same line arrives from somewhere else.

Common questions

How long should a check take?

Fifteen minutes settles most of them, one way or the other. Lines that resist fifteen minutes usually resist several hours too, because they are the ones with no trail, and "no trail after a serious look" is itself the finding.

The first search returns the answer immediately. Is that suspicious?

Not at all: plenty of quotations are real and well documented. What is worth a second look is an immediate answer that comes from many pages with identical wording and no citations, which is duplication rather than confirmation.

What if two credible references disagree?

Report the disagreement and say what each one cites. Disputed is a legitimate verdict, and it is more useful to a reader than a confident pick. The vocabulary reference works use for these states sets out what each label claims.

Do I need to check a line from a book I own?

You just did, by opening the book, that is the whole check, and it is why quoting from your own shelf is so much faster than quoting from a list. What is left is reading the paragraph around it, which our treatment of what a citation has to contain covers.