Quote context and sources analysis, one decision at a time

A dictionary of quotations looks like a list, but it is really a stack of editorial decisions. Almost all are invisible in the finished page.

Which lines got in, and what counts as a source. Whether an entry with no source is marked as such or quietly presented like the others, and what happens across editions when new evidence turns up.

Learning to read those decisions off the page is the single most useful skill for anyone who uses reference collections, because it tells you what an entry is claiming before you rely on it.

What to take away

What an entry is made of

Strip a good entry down and it has parts, each doing a different job.

The note is the part worth reading first. Scholarly editions have long used a critical apparatus to show the evidence behind a reading rather than asking the reader to trust the editor.

The quotation collections worth anything do a compressed version of the same thing. A collection with no notes anywhere is not a collection with no doubts, and it is one that decided not to show them.

Two of the standard general collections are The Oxford Dictionary of Quotations, published by Oxford University Press, and The Yale Book of Quotations, edited by Fred R. Shapiro and built from primary sources rather than from earlier dictionaries. Comparing them is useful precisely because the four decisions below were made differently in each.

The four decisions behind every entry

Selection. Somebody decided this line was worth including and thousands of others were not. Selection criteria are almost never stated, and they shape the resulting picture completely: a collection assembled for public speakers looks nothing like one assembled for students, and neither is a sample of anything.

Wording. Where variants exist, the editor picked one. Sometimes the most familiar, sometimes the earliest, sometimes a conflation. A collection that prints one wording without noting variants has hidden a real editorial act.

Sourcing standard. What the collection required before printing a name. This varies enormously, and it is the difference between a reference work and a list.

Marking. Whether an entry that failed the sourcing standard is flagged, dropped, or printed like the rest. This is the honest signal, and it is the one to look for.

Reading a collection in five minutes

The last row separates a maintained reference from a reprinted one, while long-running general collections change substantially over time as attributions are corrected and material added or dropped.

Bartlett's Familiar Quotations, first compiled by John Bartlett in 1855, is the best known English-language example, and it has been revised through many editions. That is a feature, not a flaw. Citing the collection without the edition cites nothing in particular.

Why old collections are dangerous and useful

Both at once, and for the same reason: they were compiled from earlier collections.

Dangerous, because an attribution can be traced backward through five collections and never touch a primary text. Each compiler took the previous one on trust, which is exactly the loop that makes a bad attribution look well established — the same loop that fills collections with misquotes and half-corrected names.

Useful, because they are dated printed objects. A line's presence in an edition from a given year proves it was circulating by then. Its absence from a comprehensive collection of that period is meaningful evidence about when it entered circulation.

Used as chronology rather than authority, old collections are one of the better tools available. Collecting practice changed by period, and what each generation of compilers was actually doing shapes what its books can tell you.

Borrowing the method

The practice that transformed the study of ancient texts was textual criticism: compare the witnesses, work out which readings are derived from which, and show the reasoning. A quotation entry can carry a miniature version.

That is five lines, and it converts an assertion into something a reader can check. Steps four and five only work if those labels carry a consistent meaning from one collection to the next, and a line that never gets past step four is still usable as long as it is marked unverified.

Two entries worked through

Take a familiar line. In the King James Version of 1611, 1 Timothy 6:10 reads "the love of money is the root of all evil." It is almost always quoted as "money is the root of all evil," which claims something else. An entry that delivers gives the wording it prints, the epistle, chapter and verse, the translation and its date, and a note on the common variant.

Now take a disputed attribution. "Let them eat cake" is still printed under Marie Antoinette's name, even though its earliest traceable appearance is in Rousseau's Confessions, put in the mouth of "a great princess," and no contemporary source connects it to her. The entry that delivers prints the line, marks it attributed or unverified, and points to Rousseau rather than to the queen.

Common questions

Is a specialist collection better than a general one?

Usually, on its own subject. Narrower collections tend to have editors who know the primary literature and can point inside it, where general collections inherit a great deal from predecessors.

Does an entry with a page number settle the question?

It settles that the passage is in that place, if you open it and find it. Locators are occasionally wrong, and a locator that has been copied from collection to collection may point at a page in an edition nobody has consulted in decades.

How should a collection be cited?

By title and edition, and as a collection rather than as the source of the line. Citing a quotation dictionary as though it were the primary text is the error that keeps the loop turning.

What if two respected collections disagree?

Note both and treat the line as disputed. Disagreement between careful editors nearly always means the primary evidence is thin, which is itself the useful finding.