Although no one ever taught it to you, odds are that if you solve a lot of crossword puzzles, you’re fluent in the grammar of crosswords. Most crossword enthusiasts could explain that nouns clue nouns, verbs clue verbs, and so on. They also come to know—subconsciously—that answers must be interchangeable with their clues in a sentence, even for categories too particular to have a name.
These unspoken tenets can be deceptively complex. Consider how GALORE could be clued by “aplenty,” but not by “many.” This is because, even though you might call them all adjectives, only galore and aplenty come after the noun they modify (whereas most other English adjectives come before). Additionally, simply by filling in enough puzzles, our brains can learn that multi-word crossword answers must form what linguists call a syntactic constituent—a group of words that functions together as a complete unit, the way safe and reliable does but safe and doesn’t.
The rules of crosswords are part of a rich set of conventions shared by those who solve them—and they are intimately related to the grammar of language in general. Just as toddlers develop a deep knowledge of different classes of words without being taught what a noun or a verb is, crossword solvers develop strong intuitions about what entries are possible and how they can be clued. These intuitions help you know what makes a good answer, such as OPRAH (“Founder of Harpo Productions,” in a 2019 New York Times puzzle), WHO GOES THERE (“Sentry’s query,” from 2018), and BANANA SPLIT (“Dessert served in a boat,” per a 2022 entry).
But the real sign of crossword solvers’ effortless familiarity with the rules is knowing what makes a bad answer, like SUPERB PAN (“Amazing thing to fry bacon in”) or ORDER A RYE (“What the whiskey lover would do at the bar”). Those duds hail from the work of Trip Payne, a crossword constructor and speed solver known for his absurdist “Cuckoo Crosswords.” Payne’s Cuckoo Crosswords are (deliberately) filled with entries so comically bizarre that solvers will agree that they have no place in a typical puzzle. One puzzle from 2021 contained the clue “Vanity on the set of ‘Uncut Gems.’” The answer: ADAM SANDLER’S EGO.
Whereas entries like SAFE AND would flout the basic precepts of the English language, the cuckoo entries in Cuckoo Crosswords are flouting a more subtle, if no less fundamental, rule in crosswords. In both cases, crossword solvers have learned through experience what makes a given entry acceptable or cuckoo, and also what makes some entries truly great. Although these rules are particular to crosswords, they rely on more general linguistic intuitions—hard-earned knowledge about language that’s in your brain despite the fact that you’ve likely never consciously thought about it.
Cuckoo answers such as ADAM SANDLER’S EGO are perfectly valid English phrases that could easily be substituted for their clues in a sentence, structurally the same as more acceptable entries like PANDORA’S BOX or ACHILLES’ HEEL. And yet, for reasons that can be hard to articulate, they still aren’t satisfying. Crossword constructors have a name for answers like these: ADAM SANDLER’S EGO, SUPERB PAN, and ORDER A RYE are all “green-paint entries”—as is the answer GREEN PAINT itself.
Some crossword experts argue that green-paint entries irk solvers because they are too rare in ordinary language. But you wouldn’t think twice if you encountered order a rye (“Hi there, I’d like to order a rye”) in daily life—in fact, depending on your interests, you are far more likely to hear it than a perfectly acceptable entry such as APSE. The real reason ORDER A RYE is banned from mainstream crossword society has to do with a feature of all human language called “compositionality”: the ability to combine smaller pieces of meaning into larger ones with predictable meanings. Compositionality is what lets you understand multi-word phrases—green paint is paint that is green—and sentences you have never heard before, such as “Lady Gaga splattered the crossword grid with green paint.”
Compositionality is a cornerstone of human language, letting us express an infinite number of ideas from a finite number of words. For crossword solvers, though, phrases that are purely compositional tend to be disappointing. An answer like GREEN PAINT is unsatisfying because there’s not much to it besides the meaning you get by fusing the two words together. Good multi-word answers, by contrast, are somehow more than the sum of their parts. Compare GREEN PAINT with GREEN DAY (clued in a 1996 New York Times puzzle by “Band with the Grammy-winning album ‘Dookie’”), which refers not in any way to a day that is green. Linguists call phrases like this “non-compositional” because you have to learn something about their meanings as phrases—there is no way to get the meaning of the whole just by composing the parts.
Unlike the crossword, compositionality is not black-and-white: Phrases can be fully compositional, entirely non-compositional, and everything in between. Consider GREEN TEA (“Sushi bar cupful,” per a 2003 New York Times puzzle). It is tea, and it is green (kind of). But knowing those two facts doesn’t tell you everything there is to know about green tea. It also has a particular taste, its own history, its own set of resonances. Putting green food coloring in black tea might turn it into tea that is green, but it would not make it green tea.
For crossword answers, the same logic also applies to single words. Purely compositional answers such as REBROIL (re plus broil, presumably meaning something like “broil again”) are more likely to draw solvers’ ire than words like RETURN. Even though return is, like rebroil, composed of pieces (re and turn), its meaning is not so clearly determined by the meaning of those pieces alone: Returning is not just the act of turning again.
For a crossword answer to really excel, it should, like Proust’s madeleine, evoke a particular time or place or milieu. That’s why completely compositional answers tend to fall flat: If a word or phrase is made up on the spot, it has no shot at doing that. You probably have no particular association with REBROIL or GREEN PAINT because they just don’t have lives of their own. But a favorite dessert (MOLTEN CHOCOLATE CAKE) or artist (TINA TURNER) or technology (MACBOOK PRO) might evoke a whole world of prior experiences. When we encounter these words and phrases, we remember their real-life context and even their emotional resonance.
This sense of place can add an extra layer of pleasure to phrases that might at first seem not to have any meaning or associations beyond what you get by combining the words. Take the entry SO RANDOM from a 2022 Kameron Austin Collins New York Times puzzle. By itself, it means something like “especially disordered.” But many solvers know that the phrase is deeply situated in a zeitgeist. SO RANDOM evokes an image of a Millennial, perhaps ironically or self-effacingly describing their own behavior, or perhaps even speaking in the voice of a character from Clueless. These assumptions are further supported by the clue: “Like … all over the place.” Meanwhile, an answer such as ESPECIALLY DISORDERED would have no chance of tapping into any kind of cultural context.
In recent years, a new wave of constructors and editors have broadened the cultural context that crosswords draw on for their answers. Today’s most beloved crosswords contain fewer obscure European rivers than they do celebrities, allusions, and snatches of language from all walks of life. “The more curatorial voices you have, the more the world will be reflected in the grid,” Natan Last, who constructs puzzles for The New Yorker, wrote to us. Today, those curatorial voices include Juliana Pache, whose Black Crossword has clued MENTO not as the hard candy but as “Jamaican folk music that later influenced ska and reggae,” and Nate Cardin and his colleagues, whose Queer Qrosswords often emphasize LGBTQ themes. For instance, a 2019 puzzle by Claire Rimkus and Andrew Kingsley was built around the answer AUTOSTRADDLE, the name of a queer feminist online magazine.
Crosswords reveal the intricate mental processes involved in not only combining words into phrases, but also combining letters into words. Unlike most of our experience with written language, crossword answers are written without spaces, as single entries in a dense, interconnected grid. Some great crossword answers take devilish advantage of that fact. Imagine you’re solving a puzzle and stumble across one of these two six-letter, partially filled entries: XG____ or ____DJ. Any solver could reasonably, upon seeing these bizarre letter combinations, decide that they must have gotten one of the crossing words wrong—after all, there aren’t any English words that start with xg- or end with -dj. But in these cases, the answers XGAMES (“Sports event that notably declines to drug-test its participants,” from a 2022 New York Times puzzle) and PBANDJ (“Lunchbox item, for short,” from 2003) both fit neatly into the grid.
Because language processing happens so automatically, unless you are solving a crossword, you might not ever notice that PBANDJ looks pretty weird when smashed together. Or, for that matter, that GREEN TEA isn’t just tea that is green. Perhaps that’s part of what makes crosswords so appealing: They provide a novel playground for our linguistic intuitions. Just as filling in a single letter can suddenly make obvious an answer we were struggling to recall, solving crosswords can open up a new window into the knowledge that’s already in our heads.