Contents

Introduction

What is cryptic grammar?

Anyone into cryptic crosswords will know most cryptic clues are trying to say two things at once. Take a clue like Friendly astronaut's beginning to probe alien planets (9). If we "translate" its indicator words and capitalise the fodder to more clearly show that they're just strings of letters, we see that the clue's real meaning is something like ASTRONAUT's beginning first letter to probe go into alien anagrammed PLANETS.

To grasp that cryptic reading, we have to ignore the misleading story the surface is pushing on us, and we have to give new or quirky meanings to individual words in the clue, but we don't really have to modify the grammar. In the reformatted version of the clue, the ordering of the words and the relationship between them is familiar from how English is used in real life. That word order is of course part of the meaning of the clue (to take the most obvious example, it tells us that the letter A goes into the anagram, not the other way round).

What if the clue had used "head astronaut" to indicate the A? Translating that would have given us "first letter ASTRONAUT", and we'd have needed to apply a further translation to get back to something like "first letter of ASTRONAUT" or "ASTRONAUT's first letter". For many, that sort of thing means the clue's cryptic grammar is imperfect, not so much because it's broken a rule, but because it doesn't resemble the normal patterns of English. Cryptic grammar is the idea that the grammar underpinning cryptic clues should not in itself be a puzzle – that's the job of the surface and the meanings of the individual words that make up the clue.

Does it matter?

The usual argument for paying attention to cryptic grammar is that it's fairer for the solver if they can rely on ordinary grammar and without having to guess their way towards newly invented grammatical rules like "head astronaut" can mean "head of astronaut". It's true that leaving ordinary grammar behind often leads to clues that feel more like vague guessing games than precise instructions, but it’s also true that many solvers will see no difference between "astronaut's head" and "head astronaut". It's a bit silly to suggest that looser cryptic grammar "doesn't work", as if a law of physics has been broken. Looser cryptic grammar clearly can do the job of facilitating a light-hearted word game, though there are good reasons some prefer a more rigid style.

A better argument in favour of cryptic grammar is an aesthetic one: if a clue can be read in two distinct ways and both of these conform to typical English grammar, then the setter is overcoming quite challenging odds to pull off a sort of linguistic magic trick. The crossword setter Fez has compared this to the optical illusion of a duck that looks like a rabbit (or is it the other way round?).

I just think duck-rabbits are neat, so this page is intended as a practical guide to making them, aimed at people who share this interest. When I say below that particular grammatical patterns don't work, I'm just evaluating them within the constraints of this style. This page isn't a demand that everyone writes their crossword clues like this.


About this guide

In UK broadsheet newspapers and some other contexts, the vast majority of clues use a recognisable style of cryptic grammar. The style has its origins in an influential book by an early crossword setter named Ximenes, but it also includes certain conventions and rules that evolved after that. People new to crosswords often ask if "the rules" are written down anywhere, and the truth is that many of these rules are sort of passed down without being documented in one place. Even if you read Ximenes' book you'll find some striking differences between what Ximenes said and what today's "Ximeneans" do, so I would guess most setters' grasp of cryptic grammar comes from real-life conversations or discussions on crossword-solving blogs.

This guide aims to clear things up a bit by listing different grammatical combinations of indicator and fodder and saying whether or not they're commonly seen in this shared style, which we could call Standard Cryptic Grammar. In a sense all I'm doing here is documenting or describing conventions which exist in modern UK crosswords, but I also try to explain why some things are do's and some are don't's (I also point out a handful of conventions of Standard Cryptic Grammar which, to me, don't seem to be very well-founded in ordinary grammar).

As I said, this guide is mainly based around considering different grammatical combinations of indicator and fodder (I've previously suggested these could be called fodules, and luckily this has been embraced by crossword setters everywhere).

Key

snake's tailThings no-one has a problem with
tail snakeThings usually rejected in this style
snake tailThings which some people are fine with and others have a problem with

Finally, here are a few general principles which I think we can observe in Standard Cryptic Grammar. They're all derived from the golden rule that clues should "say what they mean" (this idea was popularised by Afrit, another early crossword setter who influenced Ximenes, and basically means that clues should express their instructions in ways which can be inferred from how ordinary English works):

  • Clues should either describe what the fodder is like, narrate what the fodder is doing, or instruct the solver to do something with the fodder.
  • The cryptic reading should present the wordplay as something which is currently happening or something which has the potential to happen, not as something which happened in the past.
  • If the indicators used are the kind suggesting an action is taking place, the word order should make it clear who or what is doing that action.
  • If a clue uses "cats" to refer to those four letters, it's using "cats" to mean a string of letters, not a bunch of felines, so the cryptic grammar should treat it as singular rather than plural.
  • In general, clues where the solver can take the punctuation at face value are more elegant than clues where the solver has to pretend the punctuation isn't there.


💡 Tip: click/hover on grammatical terms marked like this for a quick definition.

Letter selections

The conventions around selecting particular letters from a word are slightly different from those around other types of wordplay, so it makes sense to deal with them in their own section.

Nouns

Nouns are words which refer to particular things or types of thing, including individuals. They might be in singular or plural form depending on whether they refer to one thing or a group of things.

Nouns like "head", "tail", "heart", "sides" and "guts" can refer to particular parts of words. I think the indicators which work best here are nouns which already refer to particular parts of physical or conceptual objects in the real world.


cat's head  serpent's tail  warrior's heart  chef's sides  dragon's guts

Patterns like these are totally grammatical ways of referring to particular parts of words, because the possessive apostrophes make clear that the particular head/tail/heart/sides/guts etc we're selecting are the ones which relate or belong to the fodder.

Note that if the indicator is singular like "cat's head", the solver is entitled to assume that they're only being asked to select the first letter from a single word. Equally, a plural indicator like "fluffy dogs' tails" is a clear and precise way to ask solvers to select the tails of multiple words. This sort of precision is a good example of how good cryptic grammar can give the solver information rather than misinformation.


head of cat  tail to serpent  heart for warrior  sides from chef  guts in dragon

These are all fine too. The prepositions "of", "to" etc do the same job as the possessive apostrophes in the previous section, by making the relationship between the indicator and the fodder clear in a way which is rooted in normal English (think of phrases like "name of school", "lyrics to the anthem", "present for mum", "ideas from scientist" and "characters in book").


head writer  tail bone  heart rate  side dish  gut feeling

We're back with "head astronaut" again. Constructions like this are probably the main source of disagreement in different approaches to cryptic grammar. These would always be avoided by people with a strict approach to cryptic grammar, because in normal speech and writing, "head writer" never means the head of a writer, "heart rate" never means the heart of a rate and so on. In fact, there aren't really any common phrases in English where "[X] [Y]" (with or without a space in the middle) refers to an X belonging to Y, so if you're asking a solver to understand these as letter selection indicators you're asking them to use an invented rule which only exists in crosswords.


cat head  serpent tail  warrior heart  chef sides  dragon guts

Many of my crossword colleagues would also reject these because there's still no apostrophe or preposition, but for me the fact the indicator is now after the fodder makes a big difference. While it's rare for English to refer to an X belonging or pertaining to some particular Y with a phrase like "[X] [Y]", there are countless such phrases in the order "[Y] [X]". For example, "city centre" means the centre of a city, and "London skyline" means the skyline belonging to London, and for me that's close enough.

There is a clear argument against the above, but people often do a bad job of communicating it because they just declare things like "party leader" doesn't mean the leader of "party", so let me quickly share an approach that helped me finally understand this one. Imagine having a conversation that's nothing to do with crosswords, then ask yourself which of the following sentences you might use if discussing a word:

  1. The first letter of PARTY is P
  2. PARTY's first letter is P
  3. The head/front/western side/leader of PARTY is P
  4. PARTY first letter is P
Obviously you'd naturally say (1) or (2), and you might even say (3) if you were feeling whimsical, but you probably wouldn't say (4), because for most people "PARTY first letter" is not a natural-sounding way to express that. So that's what people are getting at when they say "party leader" doesn't mean the leader of "party". While the "[Y] [X]" pattern sounds perfectly natural when dealing with city centres or party leaders, it doesn't sound natural if you actually try to use it to talk about parts of words.

We all know "party leader" is a completely unambiguous term for the leader of a party, but for people who favour this argument, that's sort of irrelevant: the point is that "PARTY leader" is not a phrase you'd use for the first letter of "party" in normal conversation, not because words don't literally have leaders, but because no phrase in the form "[word] [part]" ever sounds natural, even when that part is a totally non-whimsical one like a first letter. "PARTY's leader" and "leader of PARTY" don't have this problem, because they still sound natural.

To sum up, one argument says that if "cat head" can mean the head of a cat then it's not that much of a stretch to treat that phrase as a way of referring to the head of "cat" as well, and another argument says that a small stretch is still a stretch and is therefore best avoided.


redhead  cocktail  braveheart  downsides  

Here's another lot where I take a different view from many of my colleagues. There's nothing unusual about using words like "arrowhead" and "pigtail" to refer to parts of an arrow or a pig in ordinary, non-cryptic English, so for me those same words can be interpreted as references to the head of the string of letters "arrow" or the tail of the string of letters "pig". I would draw the line at treating something like "reverend" as a clue for R (the "end" of "rever"), as that's probably just trying too hard.

Interestingly enough this is one where people most people who call themselves Ximeneans are in disagreement with Ximenes himself, who argued that while "redhead" for R is unacceptable because redness doesn't have a head, "masthead" for M is defensible because masts really do have heads (presumably he'd accept "party leader" on the same grounds). I don't think anyone finds this argument very convincing these days, because we all agree that we're talking about the strings "mast" and "red", not actual mastheads or colours. We all agree a word has a head, so the only question is whether "[X]head" is a clear and fair way to refer to that head. Either it is, and any X can slot in there, or it's not, and no X can slot in there.

Adjectives

Adjectives are descriptive words which tell us what something is like, so they usually apply to or "modify" a noun (i.e. a word expressing a thing). What was that sandwich like? It was square, over-priced, delicious, unnecessary – those are all adjectives.

Nouns like "head" make it pretty clear which part of the fodder is being selected, but adjectives generally have to be used with a noun such as "part", "letter", "character" etc in order to work well.


cat's first bit  serpent's last character  warrior's central section  chef's outer points  dragon's internal parts

These all work exactly like "cat's head" etc from the previous section, except instead of nouns like "head" we have phrases like "first bit" made up of an adjective and a noun. As with the previous section, prepositional phrases such as "first bit of cat", "introduction to cat" and "opener for cat" can also be used, and a minority of people will also accept looser, apostrophe-less constructions like "cat first bit" and "serpent last character".


cat's first  serpent's last  warrior's central  chef's outer  dragon's odd

In real life, I can't really say "pass me the big" to refer to a big glass of wine (it has to be "the big glass" or "the big one"). In the same way, many adjectives can't stand alone as ways of referring to particular letters. Some phrases do work, though: ordinals i.e. words which are used to order things numerically ("first", "second", "third" etc) and a few other adjectives relating to order (e.g. "penultimate" or "last"). The key thing is whether these words can work like this in real life: I can say "I choose the first", "he was the third", "the last was the best", but it'd sound odd to say "I choose the central", "he was the outer" and "the internal was the best". Since "cat's first" and "serpent's last" function as nouns (shorthand for "cat's first [letter]", "serpent's last [letter]"), some people will accept these as "cat first" and "serpent last".


cat central  serpent inner  chef outer  dragon odd

As noted, "I choose the central", "he was the outer" and "the internal was best" all sound incomplete, so while some would accept "cat first" or "serpent last", we can't really use that pattern here. Again this is a question of what sounds normal in real life. In real life, you don't typically talk about a part or section of something by positioning an adjective in this way. The oldest member of a family is not "family oldest" etc, the lefthand side of a park is not "park left". Some people might make the argument that "January 1st" means the first of January, so "January first" could mean the first part of the word "January", but that's not very convincing as it relies on taking a very specific usage (which only really applies to dates) and expanding it to things which it wouldn't normally or naturally be used with.


first cousin  final countdown  central issue  extreme views  odd jobs

Like "head writer" and "tail bone", these don't work because these phrases wouldn't work like this in real life. "First cousin" would never mean the first piece of a cousin, "middleman" would never mean the middle part of a man, etc. I think things get even more unconvincing and forced if you try to select multiple letters in this way, e.g. if you use "odd jobs" to mean only the odd-numbered letters of "jobs".

From the perspective of cryptic grammar, the word "jobs" is a single thing, albeit a thing made up of smaller things, and normal English grammar doesn't generally let us say "[adjective] [thing]" to mean a particular part of that thing. The string "jobs" is like a street because it's a single thing made up of several smaller things. If you wanted to refer to the odd-numbered houses of a street, it'd be natural to call them the "odd houses", but you'd never use "odd street" to refer to those houses collectively. By the same logic it doesn't really make sense to use "odd jobs" to collectively refer to the odd-numbered letters in that string.


Adverbs

Adverbs generally tell us what way an action was done, so they primarily modify verbs (i.e. words expressing actions). How did you eat the sandwich? I ate it greedily, unattractively, quickly, completely – those are all adverbs.

cat primarily  serpent finally  warrior essentially  extremely unhappy  internally hired  
primarily cat  finally serpent  essentially warrior  unhappy extremely  hired internally  

While adjective-based constructions like "first day" and "central issue" are seen as bad cryptic grammar, adverb-based equivalents like "day, firstly" or "centrally, issue" are never really questioned. I find that a bit odd, because if you're talking about the first part of a day in real life then neither "firstly, day" or "day, firstly" is any more natural than "first day". I think the reason we accept the adverb forms but not the adjective ones may be that "first day" definitely looks wrong (because we know "first [X]" would normally mean an X which is first rather than the first part of X), but it's not clear that "day, firstly" would really mean anything in real life, so it's sort of an empty vessel which we can pour a special crossword-specific meaning into.

Ultimately I think "firstly", "finally" etc are probably too useful for setters to give up, but they're also a useful reminder that not everything which is widely used in Standard Cryptic Grammar is simply copied over from ordinary English grammar.

Verbs

Verbs are words which express actions. Verbs typically change their endings to express tense (when the action took place – "jump" vs "jumped") as well as person (who did the action – "I run" vs "she runs").

finish race  story concludes  begin song  song begins

Generally, verbs can't be used to point to particular letters, because none of the verbs we might be tempted to use mean quite the right thing. The verb "finish" is a good example: "finish race" might look like a way of instructing the solver to take just the last letter of "race", but "finish" doesn't mean to do only the last bit of something (try joining at the finish line and see if people think you've finished the race). The same applies to "conclude": "story concludes" doesn't work as a way to say that only the Y from "story" appears, because to conclude is to come to an end after first having had a start and middle.

I think the only set of verbs you can make an argument for here are the ones that mean to start/begin, but I know from past experience that many would reject these too. My argument is that you could tell someone to start or begin a race without requiring them to do the rest of it, so "begin song" is a reasonable way to instruct the solver to begin (and only begin) writing the word "song" in their crossword grid. We'd also understand "the song has begun" to mean that a song's earliest or opening segment has appeared, so I think "song begins" is also sufficient to say that the word "song" will begin (and only begin) by appearing the form of a S.

Single-argument wordplay indicators

This section deals with types of wordplay which do something to one bit of fodder (or multiple bits of fodder treated as one unit). Examples of this sort of wordplay are anagramming or reversing a word (or a sequence of words), or replacing it with a homophone.

Adjectives and adverbs

Adjectives and adverbs are probably the simplest way to indicate types of wordplay like anagrams and reversals. An adjective or adverb simply goes next to some fodder in order to tell the solver that it will need to undergo some sort of change on the way to the solution.


retrospective view  view retrospective  friends getting drunk  friends but drunk

You can get away using either of the first two orders, though "retrospective view" is probably neater as this follows the usual ordering for modifying a noun with an adjective in English (we say "blue door" not "door blue"). It's also fine to use variations with words like "getting" or "but", and these may help to tell more of a story in the surface reading of the clue.

Here's a good place to say that, in general, I don't think comparative/superative adjectives ("messier"/"messiest") make good cryptic indicators. An anagram might be a messy version of a word, but what makes one ordering of the letters messier than any other, let alone the messiest? You want everything in the clue to be considered and deliberate, so if your indicator includes a meaningless "-er" or "-est" which can't possibly have any specific meaning, it just looks like an edge you forgot to sand down.


upsettingly blue  blue upsettingly

The crossword setter Eccles has pointed out that using adverbs in this way doesn't really flow from the way English is used in real life, because in real life adverbs are applied to verbs ("I wept terribly") or adjectives ("terribly cute"), but not nouns (we don't say "Jeff is terribly"). Eccles' point is that in the context of cryptic grammar, all bits of fodder are nouns ("blue" is a noun meaning the string of letters B-L-U-E), so when we apply an adverb to a noun, we're doing something that's never done in real life. I don't think any setter (including Eccles) actually wants to stop using adverb patterns like this, though, so the emergency justification I'd offer is that in crosswordland, "upsettingly blue" or "blue upsettingly" always really mean something like "blue existing upsettingly" or "blue being written upsettingly", where "upsettingly" means "in an upside down or upset way". If that hangs together, there may be a thin thread that gets us back to reality from here, because I can just about imagine an old photograph with "Charlie – foolishly!" scribbled on the back.

Verbs

This section covers how to use verbs in order to tell the solver that some fodder does something or has something done to it. The cryptic grammar here often hinges on whether a particular verb is transitive or intransitive (see the summary below if you need it).

Transitive and intransitive verbs

Transitive verbs are verbs which must be done to something, e.g. the verb "push over". You can push someone over and maybe even push yourself over, but it's meaningless to just say "I pushed over". I remember this using the word "transport": you can't transport unless you're transporting something (a passenger, cargo etc).

Intransitive verbs are verbs which can't be done to anything, e.g. the verb to "fall". I can fall and you can fall, but it's meaningless to say "I fall you" or "can you fall that thing?", because nothing can "fall" another object.

Many verbs have transitive meanings in some contexts and intransitive ones in others, e.g. "flip" is transitive in "I flipped the table" but intransitive in "I flipped". Verbs like these can work either way in the cryptic grammar, and sometimes a verb that looks transitive in the surface is secretly intransitive in the cryptic grammar or vice versa.



Transitive verbs

Transitive verbs are verbs like "behead", "hate" and "insult" which must be done to something. Within cryptic grammar, transitive verbs are used to show that some fodder is falling victim to some kind of action, e.g. being anagrammed, reversed or transformed in another way.


behead beast

In constructions like this, the solver is essentially being instructed to inflict some wordplay on some fodder: "behead BEAST!". It's a perfectly natural way of expressing that, just like a recipe asking us to "drain pasta" or "chop onions".


beast behead

There are two ways you might try to justify this, but neither of them work. You can't read it as a statement that a beast is doing the act of beheading, because beheading has to be done to something, and the beast isn't doing the act of beheading to a secret second thing, it's getting beheaded itself. You might also try to read it as an instruction to the solver similar to "behead beast", but with the words back-to-front like this they no longer work as clear instructions: it's like "onions chop" or "pasta drain". You can sort of make it work if you read it as "beast (...and now behead that!)", but having to mentally swim against the current like that and insert imaginary punctuation makes things rather strained and forced.


beast gets beheaded

We just noted that "beast" can't simply "behead", because what actually happens is that it gets beheaded. Well, for that very reason, "beast gets beheaded" is a legitimate pattern. It tells us that the wordplay requires that "beast" become the victim of a beheading. It can be read as "beast gets beheaded [by the solver]".


beheaded beast  beast beheaded

These examples are actually just a variant of the [fodder + adjective] pattern we've already seen. That's because "beheaded" is actually functioning as an adjective here – it's describing the word "beast" in a way which is equivalent to a more traditional adjective like "headless". Verb-derived adjectives like this (called past participles) can be very useful when setting: if the surface includes something like "hero beheaded beast", solvers will initially read "beheaded" as a genuine past tense verb referring to some past deed by the hero, and they'll have to decipher the cryptic reading to understand that "beheaded" is actually an adjective which describes the present-day condition of the beast and has nothing to do with the word "hero". We can use the order "[indicator] [fodder]" or "[fodder] [indicator]", though the former will usually read a bit more naturally.


beheading beast  beast beheading

When dealing with transitive verbs, these patterns with -ing words don't really work, because they don't point to the right thing. If we tried to make sense of "beheading beast" as a description, it would logically refer to a beast which is doing some beheading, not a beast which is getting beheaded. It's the same basic problem as "beast behead": a thing can't simply "behead" (it has to behead something else), so it can't naturally be described as "beheading" either. You could sort of argue that "beheading beast" refers to what you get when you behead a beast, but that's a bit like saying you want your dinner to contain "chopping onions".


beast getting beheaded  beast once beheaded  beast after beheading  beast for beheading

Wordier variations like these are fair and can be useful for injecting a bit more narrative into a surface. They're all the same basic idea as "beast beheaded" and "beheaded beast", in that they refer to the version of "beast" which exists or will exist as the result of the beheading operation. "Beast for beheading" can be read like "asparagus for roasting".



Intransitive verbs

Intransitive verbs are verbs where it doesn't make sense to talk about them being done to something – they're simple actions which someone or something does on their own, like standing, falling or existing. Whereas transitive verbs let us show the fodder has passively fallen victim to something, intransitive ones let us whimsically present the fodder as actively doing something – the fodder isn't just getting anagrammed or reversed by the solver, it's anagramming itself or performing a reversal all on its own.


wolf somersaults  wolf to somersault  wolf must somersault  wolf will somersault

"Wolf somersaults" is perfectly fine as it tells us that the word "wolf" somersaults or reverses as part of the solution. The other variations here are also fine, as they tells us what "wolf" is expected or scheduled to do on the way to the answer. Incidentally, these patterns help illustrate why verbs don't really work for letter selections. The real-life action of somersaulting has a very natural figurative meaning in crosswords, but there are no existing verbs in English which can tell us that e.g. some fodder is removing all its letters except its last letter. "Wolf somersaults" gets at the idea of reversal a lot better than "wolf concludes" gets at the idea of taking the last letter.


pets somersault  pets to somersault  pets must somersault  pets will somersault

Let's say we want to reverse PETS to get STEP. "Pets somersault" doesn't really work as a way to say that. The word "pets" may look like a plural in the surface reading, but in the cryptic reading it's just the string of letters P-E-T-S, so it needs to be treated as singular. "PETS somersault" is like saying "Stranger Things frighten me". We can't just make it "pets somersaults", of course, because that'd make the surface grammar wrong instead, but this is where those "to"/"must"/"will" patterns come in handy: we can read it as "pets [multiple animals] will somersault" in the surface and "P-E-T-S [a single string] will somersault" in the cryptic grammar.

Note that "[fodder] somersaults" would be accurate if the fodder in question were genuinely plural, i.e. made up of separate words. We might use "animals somersault" or something like "a pair of animals somersault" to say that ELK and CAT reverse to make TACKLE.


wolf somersaulted

While "wolf somersaults" is fine, "wolf somersaulted" doesn't work for me. The cryptic reading of the clue should describe the answer to us, which means it should either tell us what the letters in the answer are currently doing or predict what they're about to do. Using something like "wolf somersaulted", which can only be understood as a past tense statement, is nonsensical for this. It's like a tour guide saying "Paris was the capital of France" or a maths teacher saying "2 plus 2 equalled 4".

You might ask why we can't treat "somersaulted" as a secret adjective (like the one in "beast beheaded"). The reason is that it doesn't make sense to think of an individual as being "somersaulted", because (unlike beheading) it's not an action which is done to something. Since "somersaulted" can't be an adjective, it can only be read as a statement about the past, and that doesn't work for a cryptic clue.


wolf somersaulting  somersaulting wolf

We said "beast beheading" and "beheading beast" didn't work because they'd logically refer to a beast which is doing some beheading, rather than the beast which is being beheaded. In contrast, these patterns do work, because "somersault" is a verb which the fodder can do on its own, without the need for some other victim or object. "Somersaulting wolf" and "wolf somersaulting" are therefore perfectly logical ways of referring to a wolf which is reversing. They work just like "reversing car" or "car reversing" in real life.


somersault wolf somersaulted wolf

Neither of these work, and again it's because somersaulting is the kind of action which cannot be done to something else. While you can behead a beast, you can't somersault something else, so "somersault wolf" isn't a meaningful instruction to the solver unless we redefine "somersault" as "to force to do a somersault". And while a "beheaded beast" is a beast that's been beheaded, there's no such thing as a "somersaulted wolf" or a somersaulted anything, so the patterns "wolf getting somersaulted" and "wolf once somersaulted" don't work either.

Nouns

Many crossworders would say there are three proper ways to indicate wordplay operations: 1. using an adjective or adverb to describe the fodder (e.g. "poor lamb" or "lamb, strangely"), 2. using a verb to show the fodder performs an action ("lamb running around") or 3. using a verb to instruct the solver to do an action to the fodder ("move lamb"). The idea is that in those three methods, the grammar makes the relationship between the fodder and the indicator absolutely clear.

For people who hold that adjectives, adverbs and verbs are the only acceptable ways to express wordplay operations, something like "lamb soup", which just uses the noun "soup" to suggest an anagram, is by definition unacceptable. I've spent a lot of time trying to understand the rationable here, but I still can't help feeling that nounal indicators are somewhat automatically dismissed because of what they don't do, with little thought paid to what they do do, so let's make this a longer section and consider different noun patterns based on their own merits.


anagram of nice  arrangement of Mozart  adaptation of film  version of album  sounds of animal  

These ones are sometimes begrudgingly accepted on the basis that the "of" establishes a relationship between the indicator word and its fodder by suggesting the particular arrangement/reversal/adaptation etc belongs to or is associated with the fodder. Personally, I can't see why it has to be begrudging! Clearly, "an anagram of NICE" is a not only a perfectly sensible English expression, but one you'd reach for if explaining a clue to someone you were teaching to solve crosswords: "OK, 'nice, unfortunately' is telling us we need to find an anagram of NICE". How can a phrase express a concept perfectly well in real life but apparently fail to do so within a clue?

For many the answer is that other indicators give solvers an instruction and imply a sense of ongoing movement or change, but nounal indicators do not. I think this seriously oversimplifies how all those other non-nounal indicators work. "Rearrange nice" is certainly an instruction, but what about "nice moves"? That's a whimsical invitation to observe the word "nice" moving of its own accord, and it only becomes an instruction when we infer from it that we should carry out the required move ourselves. Calling adjectival indicators instructions makes even less sense: something like "bad nice" is really a description of the end result of some wordplay and only becomes an instruction because we mentally rephrase it as "take 'nice' and rearrange it with a bad order".

It may be that instead of an instruction, "anagram of nice" just gives us a description of a category which the end result of the wordplay will belong to, but we can easily overcome this problem (if we want to) by simply exchanging that description for some letters that fit the category described. This is exactly what we do when we swap a definition like "friend of Bert" for ERNIE or swap "note" for any of the various musical notes. The fodder we'll input into the wordplay operation doesn't have to be wrapped in a verb or adjective if we acknowledge that category names can also describe the output of the wordplay operation.

I hope that's "anagram of nice" taken care of, but the next objection would probably be that while "anagram of NICE" expresses that category clearly, "arrangement of NICE" does not. If we can accept that "drunk" is a synonym of "anagrammed" and that "destroy" is a synonym of "make an anagram out of...", is it that much of a stretch to treat "arrangement" as a synonym of "anagram" itself? All of these invest ordinary words with some special meaning just for crosswords, so if we can do it for verbs, adjectives and adverbs, why not for nouns too?


role reversal  Tory U-turn  film adaptation  album version  animal sounds  

Here's where we get to the real controversy, because people who throw us a bone and allow "arrangement of Mozart" will send us straight to the doghouse again for "Mozart arrangement". Much now depends on whether you think "Do you know the famous BRITNEY SPEARS anagram?" is a sensible question. To me, "Britney Spears anagram" is a perfectly coherent phrase (though maybe not one I'd use in any context: "Everyone knows PRESBYTARIANS is a BRITNEY SPEARS anagram" might sound a little odd). Yes, it might sound more formal to describe this as "anagram of Britney Spears", but I don't think the shorter phrase is really incorrect or incomprehensible, any more than I think "Othello adaptation", "Kate Bush single" or "Ode to Joy arrangement" are incorrect or incomprehensible.

In fact, I think the comparison to those phrases brings up a useful point. Linguists sometimes talk about relational nouns, examples of which would be "sibling", "capital city" and "finale". Something (let's call it X) can only belong to one of these "relational" categories if a particular relationship exists between X and something else: any sibling has to be the sibling of someone else, any capital city has to be the capital city of a larger geographical entity, any ending is the ending of something else, etc. "Anagram" and "homophone" are all relational nouns. A word is (obviously) only an anagram if it's the anagram of another word, and a word is only a homophone if it's a homophone of another word. I think we also use "reversal" in this way within crosswordland, as in "followed by a reversal of FLOW".

I bring relational nouns up because I think there's a general rule that whenever you have a relational noun, you can talk about members of that category using an "[X] [noun]"-type pattern, at least in some contexts. A newspaper headline, for example, might say "JOHN SMITH SIBLING SPOTTED IN USA CAPITAL AFTER FRIENDS FINALE". The reason that that headline is perfectly intelligible is that the phrases it uses are all well-known ways of paraphrasing "sibling of John Smith", "capital of USA" and "finale of Friends". So I just don't buy that the relationship between "nice" and "anagram" is unclear in "nice anagram": sure, there's no preposition, but the overall structure of the phrase and our knowledge of how certain nouns work gives us a clear meaning which is available to us if we want it. So maybe we should say that nounal indicators work, but only in cases where the noun is one that implies a relationship between the fodder and the anagram.


Two-argument wordplay indicators

This section deals with types of wordplay in which two bits of fodder (or two lots of fodder treated as two units) are acted on in different ways, usually because one bit of fodder does something to the other bit of fodder. The most common examples of this are insertions, where fodder A contains or absorbs fodder B (or where B breaks into or punctures A) and deletions, where A banishes B (or where B departs A). Another example is charade indicators where one bit of fodder goes before/after/above/under/next to another bit of fodder.

Prepositions

Prepositions (words like "in", "out", "off" and "around" which are mostly used to express how two things or events relate to each other in space or time) are often used to indicate containments, deletions, charades and hidden words, as all of these types of wordplay are basically about how two bits of fodder relate to each other physically.


king out of state  king in middle of state  king within state  king off state  king underneath state  king atop state  king outside of state  king next to state

Test

after state, king  in state, king  around state, king


Verbs

We only have to consider transitive verbs here, because these types of wordplay will always involve one piece of fodder doing something to another piece of fodder.


Transitive verbs

The examples below refer to a containment expressed as A grabbing B, but the same would apply to a containment expressed as B breaking into A, and to deletions expressed as either A banishing B or B departing A.

As there are now two bits of fodder involved there are a lot more combinations to consider. The key thing to remember is to make sure that you put your two bits of fodder the right way around. Obviously "egg grabs cat" isn't a grammatically valid way to say that cat contains egg!


cat grabs egg  cat to grab egg  cat must grab egg  cat will grab egg

Patterns like "cat grabs egg" are probably the most natural way to express "two-player" wordplay operations like, as they this provide a clear statement that A does something to B. Again we can also suggest that the wordplay is scheduled or expected to happen, using something like "cat to grab egg".


cat grabbed egg

This can only be interpreted as a past-tense statement, and as I've said before, I don't think that works for cryptic clues. "Cat grabbed egg" suggests that the insertion happened at some point in the past and has now stopped happening, but a clue should really tell us what's currently happening or what's about to happen.


cat has grabbed egg  cat's grabbed egg

Whereas "cat grabbed egg" reads like a statement which used to be true and is now false, I think these are unambiguously statements about the present. Both "cat has grabbed egg" and the contracted version "cat's grabbed egg" suggest that this an insertion has taken place and can still be observed in the wordplay.


cat grabbing egg

Just as "somersaulting wolf" refers to a wolf in the process of doing a somersault, "cat grabbing egg" refers to a cat in the process of grabbing an egg, so this is another perfectly grammatical way of referring to the fodder.


cat with egg grabbed  cat egg grabbed

"Cat with egg grabbed" is similar to "cat grabbing egg", in that it refers to cat which is grabbing or has grabbed an egg. Note that although it uses "grabbed", "cat with egg grabbed" isn't really in the past tense – it's a description of the cat's current state, just like "toy with batteries included". We can also use something like "cat egg grabbed", where the "with" is implied, but this is a bit less elegant as it requires the solver to mentally insert some punctuation to decode it as something like "cat (+ egg grabbed!)" or "cat [with] egg grabbed".


egg grabbed by cat

I see these as the inversion of "cat with egg grabbed": we can read them as "egg that has been grabbed by cat" or "egg that cat is grabbing". Again, it's a description, this time of the egg, rather than a statement set in the past.


egg cat grabs  egg cat to grab  egg cat has grabbed  egg cat's grabbed  egg cat's grabbing  egg cat grabbing

These are obviously not typical English phrasings, but you can sort of justify them by pointing out that this object-subject-verb order has been used in English in the past – and there's also Yoda, of course. The crossword setter Amoeba has suggested that instead of condemning these as awkwardly phrased full sentences, we should see them as perfectly natural noun phrases pointing at particular objects – after all, we all understand that "the one I love" is a description of someone, not an attempt to say "I love the one". Even with these justifications, I think most people would say that these are less elegant than more straightforward alternatives.

In the previous section we saw how clues use transitive verbs to show that one bit of fodder contains, enters, deletes or departs a second piece of fodder. Hidden word clues are often indicated with transitive verbs too (and many containment indicators are also perfectly good hidden indicators) but the difference is that in hidden word clues one of the participants in the action is the solution itself. For example, in a clue like "Pants containing insect (5)", the fodder "pants" does the transitive action of "containing", but instead of containing another bit of fodder it contains the answer itself (ANT), as represented in the clue by the definition "insect".



Ditransitive verbs

The transitive verbs covered so far express that A is doing something to B, but there are also ditransitive verbs which express that A is doing something which involves B and C equally. The verb "give" is an example of this, because the act of giving can only really thought about if one thing (B) is being given to another thing or person (C). In grammatical terms, the sentence "Paul gives Ringo drumsticks" has an direct object (the drumsticks) and an indirect object (Ringo).

I think there are only a few ditransitive verbs which are of any use to crossword setters. For example, the verb "feed" might be used to say that one bit of fodder is fed to or inserted into another bit of fodder, or the verb "deny" might be used to say that one bit of fodder is denied or deprived of another bit of fodder. Note that in these constructions, the unseen 'A' who's doing the action involving B and C is the person solving the crossword.


feed cat egg  deny dog love  feed egg to cat  introduce egg to cat

These are direct instructions telling the solver to carry out the wordplay. Some verbs sound normal in this "feed cat egg" construction, and some don't. And for some phrases, it's possible to reword and add a preposition, but I think "deny egg to cat" would just sound odd, wouldn't it?


cat fed egg  dog denied love  egg fed to cat  egg introduced to cat  

These may look like past-tense sentences, but in terms of cryptic grammar they're actually descriptions of the cat/dog/egg, similar to "cat with egg grabbed" etc which I compared to "toy with batteries included" earlier. I find "deny" particularly useful, because on first glance solvers will probably read "dog denied love" in the sense of denying an accusation, before eventually realising it's actually "denied" in the sense of denying someone something they want or need.


dog is denied love  dog to be denied love  dog being denied love

These are all perfectly sensible too, but let's not push things too far by attempting Yoda-style sentences with ditransitive verbs... A break, give me...


Objects, statements and instructions

The patterns mentioned above cover how bits of fodder fit together with their indicators, but for the next level of cryptic grammar it's also important to think about these indicator+fodder blocks (look, I really think we should just call them fodules) fit together with other indicator+fodder blocks (OK, we don't have to). I think we can identify three basic types of indicator+fodder block, and that can help make clear whether they fit together or not.

There are what we might call objects, which basically consist of some fodder with an adjective or adverb indicator attached. These correspond to noun phrases in grammar. You can tell you're dealing with an object if it'd make sense to say "this clue features..." before it.

This clue features...  reversing car  odd bits of bus  nest with egg in it

Then there are statements, which are assertions telling us a fact about the fodder. In grammatical terms these are usually full sentences where the fodder performs a verb. You can tell you're dealing with a statement if it'd make sense to say "It's true that..." before it.

It's true that...  car reverses  bus regularly vanishes  egg goes into nest

Finally there are instructions, where you the you the solver are directly told to do some action to some fodder. In grammatical terms these are imperative sentences. You can tell you're dealing with an instruction if it'd make sense to put "you must..." before it.

The solver must...  reverse car  regularly delete bus  put egg into nest

You can press the button below to colour-code all the fodules in this guide according to whether they're objects, statements or instructions.