Finish the idiom: Jev knows the ending people forget

Given an idiom without its last word ('Be a bad apple in the ___'), does Jev give the idiom's own word, and does it follow people when they mostly give a different one?

result186 questions

Given an idiom without its last word, Jev picks the idiom's own word for 65% of 186 idioms, where 55% of people did on average (53% vs 39% for the least familiar third). On the 48 idioms where most people gave a different word, Jev sides with the idiom 38% of the time and with the crowd 27%: it knows the dictionary form better than people remember it.

least familiar53% · 39%

middle73% · 59%

most familiar69% · 66%

Jev picking itpeople giving the idiom's word

How to read this: Bars for the least, middle and most familiar thirds of idioms: how often Jev picks the idiom's own last word, next to the share of people who wrote it.

186 idioms with a completion question shown, about 100 people each; 90% interval on Jev's share [0.597, 0.71].

In short

Jev finished 65% of 186 American idioms with their standard last word, while on average 55% of people wrote that word.
The gap is widest on the least familiar idioms (53% against 39%), and where most people drifted to another word, Jev still sided with the book more often.
Jev chose from a short list containing the right word while people wrote one from scratch, which flatters Jev.

What the data shows

Finding Neverland meme: Jev finishing an idiom with its textbook last word (65%; people 55%) — Jev finishing an idiom with its textbook last word (65%; people 55%)

Jev knows the standard ending better than people remember it: 65% of idioms against 55% of people on average.
The gap is biggest for the least familiar idioms: 53% for Jev against 39% for people.
Where the crowd drifted, Jev leans to the book. On the 48 idioms where most people wrote something else, Jev picks the idiom's word 38% of the time and the crowd's favorite 27%. On "Be a happy ___", nearly half of people wrote "day" and none wrote "medium"; Jev leans to "day" there too (39% vs 31% for "medium").
On the most familiar idioms they converge (69% for Jev, 66% for people).

What it means, and what it doesn't

Jev leans toward idioms in their dictionary form, especially rare ones. If you ask it to finish or check an idiom, it will usually give the standard version, which is often not what people would say.

The list format helps Jev, and the stiff "be/get" fragments make some idioms hard to recognize for anyone, so the exact numbers overstate how differently Jev and people would complete idioms in real sentences.

Caveats

The idioms come in a stiff form. The study lists each idiom in a standard form, often with "be" or "get" in front: "Be a happy ___", "Be let off ___". Out of context, "Be a happy ___" invites "day" as naturally as "medium", so some of the crowd's "drift" is the fragment, not forgetting.
Picking is easier than writing. People wrote one word from scratch; Jev chose from the idiom's word, the other words people wrote at least twice, and "another word". Seeing the right word on a short list makes it much easier to find.
American idioms, American readers. The 870 idioms are American English, rated by about 100 US adults each. British or other idioms, and readers elsewhere, might go differently.

Jev on this experiment

Would a person find it interesting to read?: Yes76%
Does it describe you?: Yes53%
Would you have predicted it?: No51%
How fair is the comparison?: The comparison is shaky
How much should a reader rely on it?: Moderately
Which caveat matters most?: Picking is easier than writing88%

Why ask this

Idioms are phrases whose meaning isn't the sum of their words, and they wear down with use. Many people finish "a bad apple in the ___" with "bunch" rather than "barrel", or "let off ___" with "easy" rather than "lightly". Which ending people produce shows which idioms are still alive in their standard form and which have drifted.

A model trained on vast amounts of edited text might know the dictionary form better than people remember it, or it might follow the drift in how people actually talk. Which one tells you whether it writes like a style guide or like a person.

How this was done

The people and the data

The idioms and completions come from Bulkes and Tanner (2017), who normed 870 American English idioms with about 100 US adults per task. In one task, people saw each idiom with its last word missing and wrote the first word that came to mind. The project took 200 of the idioms; 186 of them have a completion question with results, and those are the ones counted here. The study also had people rate how familiar each idiom is, which is used below to split the idioms into thirds.

What Jev was asked

Finish this idiom with one word: "Be let off ___"
easy · here · work · early · easily · lightly · another word

The options were the idiom's own word and the other words at least two people wrote, plus "another word". (For this idiom, 65% of people wrote "easy" and none wrote "lightly".) Each question was asked with the options shuffled, and for "most people".

How it was measured

How often Jev's top pick is the idiom's own last word, against the share of people who wrote it, overall and for the least, middle and most familiar thirds of idioms (familiarity from the same study). Then, on the idioms where most people wrote a different word, whether Jev goes with the idiom or with the crowd.

Where these questions live

186 questions across 1 topic of the map. Each opens on the map with every question in it.

Word

Idioms186

Every question

All 186 questions behind this result, the telling ones first: the examples the analysis points to, then the ones where Jev misses, biggest gap first.

Jev’s own answer

Showing 0 of 186