He is of course right, in that simply sticking in extra words to give the clue more plausible surface (or literal) sense is unfair: the solver has to second-guess any clue that does this by ignoring the verbiage. In a sense clues like this are literally a waste of time – since the solver will have been misled to no useful purpose. Such sloppiness is rare (but alas not unknown) in crosswords from respectable sources.

As any setter with more than five minutes’ experience will know, there are times when a clue can only be made to work well if a word is inserted to link the wordplay and the definition. Take this example:

Bucket is white, we hear (4)

for PAIL. We have the definition (bucket) and the wordplay (a homophone of PALE) linked by the verb “is”. Taking Azed’s statement at face value, the clue should read

Bucket white, we hear (4)

but while the clue works cryptically, it doesn’t read nearly as well as a meaningful phrase or sentence. Inserting “is” here obviously helps the surface but this is not a case of putting in unnecessary verbiage. At the very least “is” is valid as a kind of equals sign, indicating equivalence between wordplay and definition. It helps the solver work out where the break between the definition and the wordplay occurs. That may not seem necessary in a simple clue like this, but imagine that the setter had tried to avoid the link word by phrasing the clue like this:

Bucket, we hear, white (4)

The surface reading isn’t much (or any) better than the previous example but this clue is unfair, in that we have no idea whether the answer should be PAIL or PALE. Changing this to

Bucket is, we hear, white (4)

pretty much solves the problem, although having gone this far we might as well revert to the original as it reads better.

All this is to show that link words can be harmless and sometimes helpful or even necessary, for surface and other considerations. I should note here that I have not quoted Azed to disagree with him or prove him wrong; although he successfully avoids link words in his clues most of the time (that’s admirable) he does use them occasionally himself.

Conjunctions can be useful link words too. For example

Departed or remaining? (4)

or

Departed but remaining? (4)

both are possible clues to LEFT (the question mark in both cases is just for surface sense).

Come in after fish and chips (9)

is another valid example – if somewhat unoriginal – as a clue for CARPENTER.

All the examples above have a feature in common. The only factor determining whether we have the order [WORDPLAY] [LINK WORD] [DEFINITION] or [DEFINITION] [LINK WORD] [WORDPLAY] is the surface sense; in the cryptic sense, either way round is possible. The link word functions as an equals sign does in an equation, so if A = B then B = A.

There are other link words, often used by setters, which are directional – in that it does matter which order the parts of the clue are presented in. For example:

Elgar composed for royal (5)

Not very exciting, but a safe and fair clue for REGAL. This clue tells us clearly that we need to compose the letters of ELGAR for a word meaning royal.

In this case, the clue is presented in the order [WORDPLAY] [LINK WORD] [DEFINITION] and this is correct because the link word FOR suggests that you need to solve the wordplay to get the answer. Other link words which can be used in this way are “gives”, “gets”, “produces”, “provides”, or “leads to”, and that list is by no means exhaustive. Any word or phrase which suggests that the wordplay results in the answer can be used, and the choice of these will depend on how well they fit into the surface reading.

Now let’s look at another clue:

Despondency from changing diapers (7)

Here the clue tells us that we get the answer (DESPAIR) from an anagram of DIAPERS and so is rightly presented in the order [DEFINITION] [LINK WORD] [WORDPLAY]. There are fewer alternative link words in this case and some of the ones in common use are a little debatable – such as “of” and “by”. “Of” can mean “from” in certain contexts (“out of” = “out from”), and “by” can mean “produced by” in that Hamlet is by Shakespeare, but as any student of English knows, the exact meaning of prepositions is extremely subtle and in my view, the use of both of these in clues to mean “derived from” is rather tenuous. I use them myself as a last resort, when all else fails, but don’t like them very much.

What about a clue like

A wise man for ages, possibly (4)

There’s a potential problem here. As the clue is written, it states that if you start with the answer SAGE (a wise man) and jumble it up, you get AGES as an anagram of it. In other words, we have the answer leading to the wordplay, rather than the other way round. Does this matter?

The statement the clue makes is a true one. The word SAGE does indeed need to be jumbled up to get AGES. There’s certainly enough information in the clue for us to solve it – we can simply see it as “Which word for a wise man is an anagram of AGES?” and it’s hardly difficult to proceed from there. The clue is fair, if rather untidily presented.

I have seen this sort of thing not only from the more “libertarian” setters, but also from setters who pay scrupulous attention to detail in other areas. When setting I try to avoid it as much as possible – I am something of a perfectionist – but there are times when I have an idea for a clue which I think solvers will greatly enjoy but which will only work if presented as “answer leads to wordplay” or “wordplay comes from answer”. I’ll spend a fair old while trying to get it the “right” way round, but if I can’t get that to work and I think it would be too great a shame to waste what is otherwise a good clue, I’ll run with it and hope nobody notices. I would be willing to bet that the majority of solvers won’t or if they do, they won't be the least bit bothered about it!

Other link words which are of interest are “to”, “in” and “with”. “To” is reasonable in some contexts as a “leading to” link word; for example

Settler resorted to snail mail (7)

for LETTERS, but what about

Not allowed on the phone to ring (4)

for BAND? I have a problem with this one. The use of “on the phone” in conjunction with “ring” may strike some as clever, but here the little word “to” could be misconstrued as part of the infinitive “to ring” (as in to phone). Misdirection is of course the aim of the setter but I believe this is one misdirection too far and unfair, and I groan whenever I see it (which is unfortunately quite often in some quarters).

While on this point, what about

Crazy Cosima heading off hell for leather (7)

I think this is fair, as although on the surface the link word “for” appears to be part of a common phrase, the cryptic part is unambiguous: make an anagram for COSIMA H to get a type of leather (CHAMOIS).

The use of “in” as a link word was the basis of Monk’s e-mail. I have always blithely used it, as many setters do, as a non-directional link – in other words, being unconcerned about the order of the wordplay and the definition. He suggested that as a link word, “in” implies “found in” so clues of this type should read [DEFINITION] in [WORDPLAY]. I think he’s right, and shall bear that in mind when clueing in future. The only exception is when “in” links two definitions of the answer, in which case the order doesn’t matter.

“With” as a link word has received much opprobrium in recent times. I can’t for the life of me see why. “With” is a synonym for “by means of” – you make an omelette with beaten eggs, for example. So what’s wrong with

Run off with returning Eastern European (5)

as a way to get ELOPE from a reversal of E POLE?

I’ve never seen a cogent reason why “with” is unsuitable as a link word. If this was just an idea put forward by the small minority of Internet chatterers who seem to delight in making petty criticisms without having written a single puzzle themselves, I would ignore it. But the point has been made by well-respected authorities on clueing, so I have (reluctantly) avoided using it for some time now.

Let’s move on – to link words which are, as Azed would say, verbiage. I can think of a number of examples I have seen in the national papers’ crosswords but I will invent my own here to avoid any offence. Such as

Mark’s back at school (4)

Right – Monica resettled on island (7)

for ETON and MINORCA. What possible function can “at” and “on” have as link words here? They do not suggest equality, or that one part of the clue leads to the other. They are simply stuck in there to give the clue a decent surface reading. If this sort of thing were confined to amateurs and new setters learning their trade, then I wouldn’t mind. But it does rather annoy me to see this sort of thing in the crosswords of quality papers.

I don’t want to resurrect the whole Ximenean/libertarian debate here. But let me say this: since I wrote my two pieces on Ximenean clueing a few years ago, it’s become noticeable that among some solvers and setters the term “Ximenean” has taken on a somewhat negative tone, rather in the way that “political correctness” and “Health and Safety” now tend to refer to the ridiculously over-zealous pursuit of what were originally well-meaning causes. “Ximenean” is now used by some to evoke the idea of pedantic old buffers who want to put crosswords in a glass case in a museum. In some ways this is understandable – for example not so long ago I read a discussion in which the classic

Bust down reason? (9)

for BRAINWASH was declared by some to be no good after all, as the definition (which is the whole clue, as is the wordplay) doesn’t have an object. Well, maybe that’s technically right, but the clue’s still a bloody good one and it’s this kind of nit-picking that puts people off the whole Ximenean caboodle. That is of course a great shame – Ximenes himself provided by far the best guidelines for what makes for good cluemanship without being prescriptive or preachy, and it’s hardly his fault that a few people have taken the original ideas too far.

Without a doubt the greatest contribution Ximenes made was to ensure that solving was enjoyable and satisfying. He wasn’t laying down immutable laws or establishing some sort of “cruciverbal correctness” – quite the opposite in fact. He was making sure that solving puzzles was not a slog on account of the setter being sloppy.

The examples I’ve given of bad clues for ETON and MINORCA are simple enough that the redundant link words are unlikely to cause many problems. In more complex clues, though, unnecessary words like this could waste a considerable amount of the solver’s time without offering any reward at all. I’ll admit that when I (fortunately rarely) encounter this and other forms of sloppiness in the nationals, my reaction is generally “why the hell do I bother?” By which I mean why do I bother taking so much time – sometimes to the point of spending a whole morning on just two clues – to do my best to make my clues as neat and tidy as possible? I’m not saying all my clues are good – far from it. I’ve written plenty of so-so ones and some clunkers too, but I like to think that those clunkers were at least fair clunkers. It’s not just me of course – the vast majority of setters take pride in their work and will wrestle with a recalcitrant clue until they’ve got it absolutely right.

Let’s leave the rant there and sum up. In an ideal world, there wouldn’t be any need for link words. In practice, many clues do require some sort of link between wordplay and answer, and there are quite a few words in the English language which perform this task neatly. In a number of cases, these words are directional in the sense of [ANSWER] comes from [WORDPLAY] or [WORDPLAY] leads to [ANSWER]. Ignoring the directional nature of such link words does rather detract from the elegance and neatness of the clue, but sometimes the setter can get away with this if the clue is in other respects a very good one. It is wise, though, to take to heart the quote from Azed at the start of this piece and not insert redundant words into the clue just to make it flow better. Even if many solvers don’t notice (or care about) these last points, the setter's aim is to use the versatile and rich English language to provide cryptic instructions to the solver, and the more accurate those instructions are, the better all round!