Jeremy Butterfield Editorial

Making words work for you

Leave a comment

Comma, comma, comma, comma, comma, Chameleon. How to use commas (2). Pompeo and commas and CMOS.

3-minute read

In the wake of Secretary Pompeo’s edicts about punctuation, which hit the U.S. headlines in late September this year, I blogged about the conflict between commas as what I will call “syntactic boundary markers” and as pauses in speech. I suggested that commas are art, not science. By which I mean that there are several circumstances in which most authorities agree they are optional. Inserting or omitting them thus becomes a matter of personal style, not of blind rulebook-following .

The State Department circulated emails with examples of good and bad comma use. An extract from these Pompean edicts illustrates my contention perfectly. The Chicago Manual of Style (henceforth CMOS) 6.26 gives the example below and the State Department emails lifted it verbatim – except that they added the comma after and, suggesting it be removed.

Burton examined the documents for over an hour, and, if Smedley had not intervened, the forgery would have been revealed.

First, it’s worth noting that CMOS itself says this: “When a dependent clause intervenes between two other clauses joined by a coordinating conjunction, causing the coordinating and subordinating conjunctions to appear next to each other (e.g., and ifbut if), the conjunctions need not be separated by a comma.” [Underlining and emboldening mine].

“Need not” does not mean “must not” or “should not”.

Second, at the end of 6.26 CMOS says, “Strictly speaking, it would not be wrong to add a comma between the conjunctions in any of the examples above.”

CMOS is thus indulging in a sort of now-you-see it, now-you-don’t disclaimer.

Moreover, I suspect that to understand the reasoning behind the veto on commas in those circumstances could strain even the most nitpicky State Department staff because of the terminology involved. But, in case it helps you, gentle reader, here goes. (There are links to Englicious’s helpful glossary for each term.)

A: Burton examined the documents for over an hour, = MAIN CLAUSE



D: Smedley had not intervened, = DEPENDENT (or SUBORDINATE) CLAUSE

d: the forgery would have been revealed. = SECOND MAIN CLAUSE

Whether you insert that comma or not, IMHO, depends on how comma-friendly or comma-averse you are.

I would retain it but accept that others will consider it fussy.

My argument would be that writing the sentence comma-less thus

Burton examined the documents for over an hour, and if Smedley had not intervened, the forgery would have been revealed.

seems to me incomplete. And it seems so for a reason that CMOS also admits: “Such usage, which would extend the logic of commas in pairs, (see 6.17) may be preferred in certain cases for emphasis or clarity.”

A subtle argument could be made that that comma is dispensable in the sentence as it currently stands but would become necessary if the dependent clause were extended, for example like this:

Burton examined the documents for over an hour, and if Smedley had not intervened, the forgery would have been revealed.

Burton examined the documents for over an hour, and, if Smedley had not intervened so excitedly that he seemed to be on the point of blowing a gasket, the forgery would have been revealed.

In academic writing, where long sentences are the order of the day and the authors themselves often get trapped in the maze of their own verbiage, I tend to insert such commas to break up the flow and provide balance to sentences. But the length of that dependent clause can have an impact, as also might the weight and balance of the surrounding clauses, as I hope the comma-laden example above suggests.

In contrast, the following example is from a work whose prose style has been described as “needlessly obscure.” That said, and despite what I say in the previous paragraph, I would not insert a comma after the highlighted if.

If, indeed, as Fallon, Quilligan, and Franke argue, Milton’s Paradise Lost eschews the sacramental innocence of the sign that has been miraculously transformed into a sacred object, and if the poem’s central epistemological claim is to the internal processes of interpretatively spiritual (that is subjective) truth, then the poem’s uniformity of vision and tactile materiality lend these processes real, indeed tangible, substance.

(It’s only 62 words long but feels wearisomely longer to me.)

And the reason I wouldn’t is that that if follows on clearly and logically from the If that introduces the whole sentence. It is a discoursal if being used to construct an argument, in that way that connotes “let us suppose this proposition to be true, and I too am doing so for the sake of my argument”. It is not the hypothetical if of the CMOS examples, in which something might or might not have happened.

Over such minutiae – now there’s a word I can never quite decide how to pronounce, but the link shows I am not alone – do we editors cavil. Perhaps it really is time to get out more.

But before I put on my coat, here’s a question for any editors “out there”. Would you leave the emboldened comma in this (authentic) sentence or remove it?

The upregulation of myocardial beta-1 receptors has been shown to re-sensitize the myocardium to adrenergic stimulation with dobutamine and, if a similar upregulation of sinoatrial beta-1 receptors took place, may partially or fully restore chronotropic competence.



Pompeo and commas and CMOS. How to use commas (1).

5-minute read
(Less for you speed-readers out there: well done, you!)

James Thurber was once asked why there was a comma in the sentence “After dinner, the men went into the living room.” He replied that it had been added by Harold Ross, the New Yorker editor, and that the comma was Ross’s way of giving the men time to push back their chairs and stand up.

A stickler for commas

A fortnight or so ago, Mike Pompeo, the U.S. Secretary of State (for Brits, sort of the equivalent of the Foreign Secretary) was in the news because of emails circulating in his bailiwick urging, at his behest, careful use of commas: “The Secretary has underscored the need for appropriate use of commas…”

The hacks pounced on it gleefully, like flies round you know what; here was another golden opportunity to mock a member of an administration most of them detest.

Besides, his pedantry was piquant grist to their mill of slagging a President who rides roughshod over basic spelling and punctuation rules. [I’m not terribly keen on “piquant grist”, but if you must, you must; nor on the rest of the sentence, really. Sigh. Ed.]

They claimed to find it baffling, if not downright ridiculous, or[,] at the very least[,] highly suspect, that a man concerned with weighty matters of state could be bothered about a piddling little convention. The patronizing tone of mock-incredulity abounded, as displayed here.

But hang on a moment.

Let’s leave the toxic politics1 of all this aside for a while… [note, two words, a while].

Journalists are (allegedly) literate. Those Americans who write (rather than pontificate orally) have their very own style guide, the AP Stylebook. Most serious printed media similarly have their individual style guides (e.g., in Britain, The Times, The Telegraph, The Economist, The Guardian/Observer).

Such style guides don’t provide comprehensive guidance on how to use commas – presumably because any competent journalist is presumed to already know (a dangerous presumption these days, when much online news seems to be written by novices or interns whose grasp of the finer points of English can be hazy). Journalists who belittle attempts to help State Department officials punctuate “better” could be considered a mite disingenuous.

What’s wrong with good, old-fashioned rules?

What could be wrong with proffering advice to drafters struggling with the minutiae of comma use?

In the punctuation pecking order, commas are the most underrated and overlooked mark; yet[,] they are the most versatile and useful – and, surely, the most frequent.

Being the most versatile, they are also the most complex. As illustration, for example, the excellent Penguin How to Punctate devotes a generous 54 pages to them[,] compared with the 15 it devotes to the full stop[,] and the 16 to the colon.

However, it’s easier to give simple rules than to say, “Well in some cases do this, but in others do that, it’s all a matter of editorial judgement”[,] as the Penguin book does. Otherwise, who knows where we might end up? And simple rules is what the emails circulating in the State Department enjoin.

Or are they simple?

Hardly a science

The problem is, wielding commas is an art, not a science.

The emails cite the Chicago Manual of Style (CMOS) as their bible[,] and refer to specific rules. [I put a comma there, and Grammarly and CMOS don’t like it, but I do. But then, I have been accused of over-commaing.]

The following type of comma was approved:

  1. CMOS 22 states: “When independent clauses are joined by andbutorsoyet, or any other coordinating conjunction, a comma usually precedes the conjunction.”

So far, so good, but it then goes on to say, “If the clauses are very short and closely connected, the comma may be omitted (as in the last two examples) unless the clauses are part of a series.”

Rules within rules and wheels within wheels

So, we have Rule A, with exception B, which has its own exception C.

Simples, egh?

The State Department example, lifted straight from CMOS and showing appropriate use, was: a) We activated the alarm, but the intruder was already inside.

Is that first clause “very short”, at a mere four words?

Seemingly not, in this instance.

Yet the examples given by CMOS for exception B have comma-less first clauses with that exact same number of words.

b) Electra played the guitar and Tambora sang.

c) Raise your right hand and repeat after me.

Both have four words, like that clause which is closed off by a comma We activated the alarm.

Which raises the question, what is “very short”?

Perhaps we should [,] therefore [,] resort to the clauses being “closely connected”. But how does one define, let alone measure, that “connectedness”?

I repeat, commas are an art, not a science, and this would-be rule only highlights that fact.

Pause (possibly for thought)

I think most people (editors) would happily accept the need for a comma in example a) (We activated the alarm, but the intruder was already inside).

Similarly, a comma in b) (Electra played the guitar and Tambora sang) might seem OTT to many. But could one, hand on heart, say it was plain wrong?

And the same applies to c) (Raise your right hand and repeat after me).

If you base where to put commas purely on grammatical function/syntax, you ignore one of their key functions, namely, to indicate pauses, that is, treating what was written as if it is to be spoken; and [,] to provide emphasis.

For example, if spoken, Raise your right hand and repeat after me would sound perfunctory and formulaic, whereas Raise your right hand, and repeat after me arguably matches the gravity of the occasion.

The Penguin guide suggests several tests to decide where to insert a comma. One of them is “If in doubt about a comma, apply the ‘pause test’. Say the sentence to yourself, and if you hear a pause, put in a comma …”

Fine, as far as it goes, but I might “hear” a pause [,] and you might not, or vice versa.

And it seems that in the past [,] writers “heard” pauses more often than we do.

The twelve years, continued Mrs Dean,  following that dismal period,1  were the happiest of my life: my greatest troubles in their passage rose from our little lady’s trifling illnesses, which she had to experience in common with all children, rich and poor. For the rest, after the first six months,2 she grew like a larch,3and could walk and talk in her own way,4 before the heath blossomed a second time over.

Cap. 18, Wuthering Heights, Emily Brontë

The first marked comma separates its subject – The twelve years – from its verb – were – in a way not nowadays allowed (although it is, in my experience, quite common in academic writing, mainly due to long-windedness).

The second, separating a prepositional phrase saying “when” from the main clause, could nowadays easily be left out.

The third clearly contravenes CMOS 6.23: “A comma is not normally used to separate a two-part compound predicate joined by a coordinating conjunction  (A compound predicate occurs when a subject that is shared by two or more clauses is not repeated after the first clause.)” [Emphasis mine].

And the fourth, separating a following subordinate clause from its main clause, is not nowadays generally considered necessary.

Could anyone say they are wrong? Outmoded, possibly, but effective.

“If you take hyphens seriously, you will surely go mad” (in my case, my friends say[,] that boat sailed long ago) is a quotation doing the rounds, possibly uttered by a frustrated lexicographer.

The thought applies equally to commas.

1 In a comment on one of the reports about Pompeo’s punctiliousness, someone said: ”It’s so comforting that a person suffering from OCD is one of the adults in the room with the person with malignant narcissistic personality disorder.”


Twitter never fails to disappoint. Or should that be ‘never disappoints’?

While reading something online the other day I came across the phrase Twitter never fails to disappoint. The context made it clear that the meaning intended was ‘Twitter never disappoints’. This is the exact opposite of the logical reading that ‘Twitter always disappoints.’

That example reminded me of one from years ago. A tourist brochure for a seaside resort promised something along the lines of ‘A visit to X-on-Sea never fails to disappoint.

And then, slap my thigh, today, when I was checking out restaurants for my partner’s birthday, what did I come across but this glowing recommendation: I’ve been going to X Bistro in Y since it opened, which was not yesterday, and I can safely say that their food has never failed to disappoint?

(Which shows that the phrase is not a completely frozen idiom, because it allows past tense.)

What is going on that makes a structure mean the opposite of what the speaker intended? And how do other speakers manage to extract the correct meaning? The discussion on the English StackExchange site shows that the phrase can certainly cause confusion.

Multiple negations cause problems

It’s all to do with the number of negations, and how the human brain goes into meltdown when trying to process too many. Having two negations might be the limit to easy intelligibility.

Such negations can be explicit (not, no, nobody, never, etc.) or they can be implicit (fail, ignore, avoid, etc.). If we analyse our phrase in terms of negation, we’ll find three:
• to fail to do something is not to do it = negation1 (explicit)
never adds negation2 (explicit)
disappoint adds negation3 (implicit)
(Disappoint is implicitly negative since it means ‘not to live up to expectations’.)

Logically, to never fail to do something means ‘to always do’ it. ‘Twitter never fails to disappoint’ therefore means ‘Twitter always disappoints.’

But the example which caught my eye was intended to mean the exact opposite. It reads like a conflation of ‘never fails to please’ or some appropriate positive verb, and ‘never disappoints’.

Notice how the reply at the top uses the logical meaning to rebut the positive but mistaken one under the image of the woman eating.

Not a unique case

Twitter never fails to disappoint is hardly a unique case of a phrase meaning the opposite of what the speaker intends. Another well-known and well-embedded example is the ‘It is impossible/difficult/hard to underestimate’ structure, where, logically, overestimate is meant, e.g. ‘It would be impossible to underestimate its [sc. Ulysses’] influence; the novel was never quite the same again.’ The logical meaning is ‘its influence cannot be overestimated’ i.e. exaggerated.

But there we only have two negatives rather than three: one explicit – impossible – and one implicit negative in overestimate, because to overestimate is to produce an incorrect estimate.

But let’s get back to never fail to do.

never + fail + to what?

In theory, in the sense of always doing it, you could never fail to do practically anything, for example, I never fail to eat Marmite at breakfast.
However, our old friend collocation kicks in strongly here. The string never + FAIL [sloped capitals mean ‘in all forms’] + to-infinitive very often goes with events and emotions that can be classified broadly as either positive (entertain, amuse, please, delight, inspire, etc.) or intense (impress, amaze, surprise, etc.), or a mixture of the two.

Even an apparently neutral verb such as make goes with positive verbs, e.g. MAKE + me/us/people, etc. + laugh/smile/giggle/chuckle (though whether that is, in any case, a feature of make, rather than of the entire phrase, is impossible to tell).

‘You’ve got to ac-cent-tchu-ate the positive…’

In fact, the top five collocations by frequency of never + FAIL + to-inf are (in my corpus, OEC Monitor Corpus April 2018) impress, amaze, make, deliver, disappoint.

If, as I have suggested above, the overall ‘profile’ of never fail to is positive, then speakers view never fail to disappoint as positive, despite its meaning the opposite. They take the whole as a ready-made, rather than analysing its meaning.
Moreover, it is possibly one of those phrases where the presence/absence of a negative makes little difference to the meaning. As Language Log pointed out, fail to miss behaves like that: the meaning is the same whether you say miss or fail to miss. Similarly, whether you say never DISAPPOINT or never FAIL to disappoint, the meaning is the same.

The corpus I consulted contains 226 examples of never FAIL to disappoint. In a random sample of 50, 45 showed the illogical meaning (= ‘never DISAPPOINT’)

We were rewarded with our choice of route as the New Zealand scenery never fails to disappoint. (= ‘never disappoints’)

If I’m going to drop $20 on a couple of made-to-order burgers, fries and a soda, there are a few Portillo’s close to here which are similarly priced but never fail to disappoint (= ‘never disappoint’) …The staff here is on point. Honestly, they can’t do enough for you.

A mere five (10%) exemplified the logical surface reading, meaning ‘always succeed in disappointing’.

For example, in this about the chronically inept Royal Bank of Scotland (RBS):

Lord Oakeshott, a leading LibDem peer, said: ‘RBS never fails to disappoint. Taxpayers poured £45 billion but it is a zombie bank, shrinking instead of lending.’

Similarly, this investigator of financial shenanigans:

My investigations often lead me into contact with British law enforcement and regulators and they never fail to disappoint me by their incompetence and lack of professionalism.

All in all, then, it would seem that the apparently negative ‘never FAIL to disappoint’ is well established as meaning the opposite of what it seems to mean, and as positive in intent.

We interpret it as positive, I submit, because a) we are now well used to a range of constructions that mean the opposite of what they are intended to mean and b) multiple negatives cannot be processed and, in the absence of evidence to the contrary, lead to a positive or affirmative interpretation.

It could also be significant that the opposite – always FAIL to – does not collocate with the same verbs as never FAIL to. There is a solitary example of always FAIL to disappoint:

The beauty of Smashing Pumpkins is that every album is drastically different from each other. I’m eager for this release, Billy Corgan has always failed to disappoint me.

This general phenomenon of muddled negation is described by Language Log as ‘misnegation’ or ‘overnegation.’ Here is a link to a very long list of examples.

And here is an utterly mind-boggling example, courtesy of LL:

These contrasts don’t mean that Bush was without blemish: As Meacham notes, there were political misjudgments and calculated concessions to ambition on the long path to power. Nor does it mean that Trump doesn’t lack his own kind of strengths, not the least of which is his loudly declared indifference to elite opinion.

The fact is, we are able to interpret these car-crash negatives correctly and extract the meaning the speaker intended.

As humble proof of that, I stared at this oft-cited canonical example for ages before I realised what was wrong: ‘No head injury is too trivial to ignore.’ Like you, gentle reader, I understood what it meant without needing to analyse it, but it should, obviously – D’oh! – on reflection, be rephrased as ‘No matter how trivial your head injury seems, we will not ignore it’ or ‘No head injury is too trivial to be attended to.’ Again, it’s a case of that triple negation; no head injury1; is too trivial to X2 (= ‘is so trivial that it will not be Xed)’; be ignored3 (negative = ‘will not be attended to’).

Watch out for this kind of phrase. There are never too many of them not to fail to ignore.


Objections to ‘to be done with something.’ Uniquely American? I’m done with the topic, anyway.

(four-minute read)

Here’s a language issue that’s new to me.

The other day on Twitter @The_GrammarGeek asked:

‘There’s an opinion out there that it’s wrong to use “done” to mean “finished,” as in, “I’m done with my homework.” But this use of “done” has been widely used since the 15th century. Any idea/when where the false rule originated?’

Another tweep, Karen Conlin (thanks, Karen!) then tweeted that this issue is not mentioned in my edition of Fowler (4th edn., 2015), and asked if I could shed any light on it. Here goes, then…

If you’re in a hurry…

  • That highly specific use (= ‘to have finished, completed + NOUN’) seems to be mainly U.S.
  • So, strictures against it have no reason to appear in Br.E. manuals.
  • That specific use is 18th century onwards, rather than 15th.
  • According to M-W’s Concise Dict. of English Usage, objections to it were first raised in 1917, with no obvious justification.

If you’ve got longer…

Here’s my two pennies’ worth.

First, to use ‘done’ in exactly that construction, namely, HAVE + done + with + NOUN and with that precise meaning (= ‘to have completed’), is not something I personally would say (is not part of my ‘idiolect’), and – I’m speculating here – is not something most Br.E. speakers would say either. (Looking for evidence in do, one of the most common verbs in the language, could be a Herculean, not to say Sisyphean, enterprise!)

However, I might say  ‘I’m done with blogging’, using the pattern to be done with + –ing form (verbal noun), but I think that is a slightly different meaning (‘I will never do it again’ = ‘I’m through with blogging’).

And I would also write, though probably not say, the standard phrase ‘let’s tell him and be done with it’.

If the above claim is true, then there is no reason why a fatwa against the use should exist in Br.E. usage manuals. I’ve checked in all three previous editions of Fowler, and the issue has not been treated. My additions and amendments were based on notes kept over several years about issues that had struck me, and this was not one of them.

Second, what exactly is this use, and where does it come from?

What can the OED can tell us?

Previous edition

The previous edition (1989) makes it a second sub-sense under the more general, somewhat undifferentiated rubric of

8. (In pa. pple. and perf. tenses.) To accomplish, complete, finish, bring to a conclusion. to be done, to be at an end.’

The sub-sense is headed

b. to be done is used of the agent instead of ‘to have done’, in expressing state rather than action. (Chiefly IrishSc.U.S., and dial.)

That geographical information in brackets is important.

The first example given dates to 1766, from T. Amory’s Life of John Buncle II. x. 365

I was done with love for ever.

(Amory, btw, grew up in Ireland.)

The second citation, however, is from Thomas Jefferson: 1771 T. Jefferson Let. T. Adams in Harper’s Mag. No. 482. 206

One farther favor and I am done.

Current edition

The current edition (3rd edn., March 2014) is more nuanced. It puts that Life of John Buncle quotation (I was done with love for ever) at the head of a category (10. a. (b)) captioned thus:

‘Of a person: to be at the end of one’s dealings with, to have no further truck with; = sense10b(b).’

In other words, it makes it equivalent to ‘to have done with something/someone’ as in Shakespeare’s Do what thou wilt for I haue done with thee, and as the earlier edition also did.

On that analysis, the Buncle quote could have been I had done with love forever.

The meaning that is truly the one at issue, I think, is now lexicographed as follows (underlining mine):

10. a. (c) Of a person or other agent of action: to be at the end of what one is doing, to be finished. Also with complement expressing the action being finished. Now chiefly U.S.’

That note ‘chiefly U.S.’ chimes with Karen’s hunch that the use is more U.S. than British and is substantiated by the citations the OED chose:

The Jefferson quote heads that category, and the other examples are, with one exception, U.S.:

1876   H. B. Smith in Life (1881) 404   After this is done I am done.

1879   Literary World 6 Dec. 400/1   The mills of the gods are not yet done grinding.

1883   Cent. Mag. 25 767/1 ‘ twenty-four thousand dollars! Are you all done?’ He scanned the crowd.

1971   M. B. Powell & G. Higman Finite Simple Groups i. 5   Since g is arbitrary, we are done [i.e. we have completed the proof].

1981   J. Blume Tiger Eyes (1982) xxi. 87 ‘Davey..are you almost done?’ Jane calls, knocking on the bathroom door.

2000   A. Hagy Keeneland 242   You are full of total dog shit. I’m done putting up with you.

Note the examples with the –ing form, which I noted earlier that I would use. I might also say, similarly to the 1981 example above, ‘Are you quite done!’ as a retort to someone, for example, who was being rude or offensive at length.

Quick statistical note

A trawl in the Feb. 2018 ‘Monitor Corpus’ of the Oxford English Corpus for the string BE + done + with + –ing form retrieves 1,029 examples. Almost half are of unknown source, but of those whose source is known 265 are U.S., 65 British, e.g.

DANIEL Craig is said to be done with playing Bond, but producers are willing to do the impossible to keep the superstar happy.

As he’s mentioned in the example above, I couldn’t resist the temptation to add a variant of the almost legendary image from Casino Royale to add pep to a potentially dry topic.

An earlier version of the corpus (2014) shows a not dissimilar ratio.

Yes, but what about the prohibition against?

As I mentioned at the beginning, the M-W dict.’s earliest note is from 1917.  The M-W entry also notes the Heritage Usage Panel in 1969 47 percent disapproved of it, suggesting that it was a rule that had been forced on many of them.

Is it still being trotted out/bandied about? If so, please let me know where.


“that” or “which”? Using “which” in restrictive or defining relative clauses (2/∞)


A young scholar struggling with the “which/that” distinction.

There are one or two loose ends to tie up from the previous blog on this topic, before I move on.

1.1 which or that in defining clauses: with indefinite pronouns

First, it may be helpful in distinguishing restrictive or defining relative clauses from non-restrictive/non-defining ones to note the following: they often associate with specific words or kinds of words such as something/nothing/anything/everything whose very meaning suggests that any relative clause following them has to be defining, since those words themselves are indefinite (actually, “indefinite pronouns”).

a) Elevating the usually ordinary exercise of changing level to such a dramatic experience is something that Libeskind [sc. the famous architect] relishes.
Architecture Week, 2004.

b) The public showing of something which is so private and particular is immediately startling.
Art Throb, 2004.

c) Punctuation serves a valuable purpose – it helps to convey meaning more precisely and anything which erodes the precision of the English language is to be deplored.
Telegraph, 2014.

d) For Samsung, anything that could help it look better in the eyes of U.S. Federal Court Judges is probably a good move, although in this case it may not help much.
The Mac Observer, 2014

While which can be used after these words, as illustrated, it is very much a minority trend: in the case of something that/which, for example, a little less than 10 per cent of all cases.

1.2 with determiners and predeterminers

Another class of words often associating with defining relative clauses is “determiners” and “predeterminers” such as some, any, many, most, several, other, all, both, each, every, little, few, etc., e.g:

e) Icelandic law prevents the importing of new strains to prevent disease: any horse which leaves Iceland can never return. Open Country. BBC, Radio 4.

f) But there are some things that all can understandGuardian Unlimited, 2004

g) There is no herbicide that controls all plants. UNL Neb Guides, 2002.

Most of these examples show that/which as the subject of its clause. Where it is the object, as in a) and f), it could just as easily have been left out altogether, as often happens in speech, e.g.,

f) But there are some things that all can understand.

Clauses of the type, …all can understand…, from which the relative pronoun is dropped, are what is known in grammar as  “contact clauses” and are very common in spoken language.

2 non-defining or non-restrictive clauses

As mentioned in the earlier blog, the information they contain can be omitted. Putting it another way, they are almost like an aside. That is why such clauses are conventionally and correctly enclosed in commas if they come in the middle of a sentence, or are preceded by a comma if they are the last clause in a sentence. Fowler (1926)  noted that a non-defining clause “gives a reason…or adds a new fact.”

The example given in the earlier blog was “I saw Kylie Minogue, who was staying at the hotel opposite.” Even if such clauses are omitted, the sentence will still make sense (though it will, obviously, convey less information): “I saw Kylie Minogue” makes perfect sense.

In that earlier blog, there were adjoining sentences, each with a non-defining clause: “They then wanted me to review the proofs, which the publisher had had proofread. Excellently proofread they were, too, complete with a useful, comprehensive list from the proofreader, which explained their decision on style issues such as the treatment of names and titles.”

Without those non-defining clauses, each sentence still works: “They then wanted me to review the proofs. Excellently proofread they were, too, complete with a useful, comprehensive list from the proofreader.” As Fowler, noted, the clauses add “a new fact”.

3 the rule – who enforces it?

To claim that “It is a rule that ‘that’ must be used to introduce a defining relative clause’” draws attention to the ambiguity, or at least polysemy, of the word “rule”. The Oxford Online Dictionary defines two relevant senses:

  1. One of a set of explicit or understood regulations or principles governing conduct or procedure within a particular area of activity.
    ‘the rules of cricket’
    1.1 A principle that operates within a particular sphere of knowledge, describing or prescribing what is possible or allowable. [my underlining]
    ‘the rules of grammar’

The “rule” that that has to be used clearly falls largely under definition 1 above.

It is a “regulation” or “principle” “governing conduct” within a particular “area”.

In this case, the “area” is written, edited English. However, the proponents of the rule would wish to assimilate it to definition 1.1.

Clearly “which” in a defining relative clause is both possible and allowable. But the usage absolutists would wish it weren’t, and certainly consider it undesirable. Their fatwa, however, is not like a genuine rule of grammar, such as “A clause in the English declarative mood has the subject followed by the verb.”

4 Who says you have to use “that”?

4. 1 Many people. For British English, the style guides of choice are The Guardian/Observer, The Telegraph, The Times and The Economist.
4.1.1 The Guardian endorses the distinction; as does the Telegraph Style Book, but with lamentable punctuation, in what looks suspiciously similar to The Economist’s perfectly punctuated dictum. The Telegraph has “which and that: which informs that defines. This is the house that Jack built, but: This house, which Jack built is now falling down.”

The Telegraph thus misses out the essential comma closing off the non-restricting clause.

4.1.2 The Economist correctly has: “which and that Which informs, that defines. This is the house that Jack built. But This house, which Jack built, is now falling down. Americans tend to be fussy about making a distinction between which and that. Good writers of British English are less fastidious. (“We have left undone those things which we ought to have done.”).”

(IMHO, for “fastidious” read “anal”.)

However, the Economist style guide occasionally (inevitably?) breaks its own rules, e.g. The Arabic alphabet has several consonants which have no exact equivalents in English (note that “determiner” several, as mentioned earlier).

4.2 For U.S. English,

4.2.1 Garner is dogmatical and absolutist on the matter:

“Legal writers who fail to distinguish restrictive from nonrestrictive clauses—and especially that from which—risk their credibility with careful readers. It’s therefore worthwhile to learn the difference so well that, when writing, you use the correct form automatically.”

Such cut-and-driedness is a reflection, presumably, of the need for absolute clarity and unambiguousness in legal writing.

4.2.2 Chicago is more nuanced: “Although which can be substituted for that in a restrictive clause (a common practice in British English), many writers preserve the distinction between restrictive that (with no commas) and nonrestrictive which (with commas). The APA (American Psychological Association) prefers writers to observe the distinction, and the AP style guide imposes it too.

Fowler and his crystal ball

In his 1926 Dictionary of Modern English Usage, Henry Fowler included lengthy entries on that as relative pronoun and which((that)(who. In the entry for that, he distinguishes the two kinds of clause and assigns them what he considers their appropriate pronoun.

His learned, measured style is perhaps somewhat alien to modern sensibilities and is possibly easier to follow if read aloud:

“The two kinds of relative clause, to one of which that & to the other of which which is appropriate, are the defining & the non-defining; & if writers would agree to regard that as the defining relative pronoun, and which as the non-defining, there would be much gain both in lucidity & in ease. Some there are who follow this principle now ; but it would be idle to pretend it is the practice either of most or of the best writers.”

While Fowler expressed a velleity, it seems that the combined weight of usage guides, not to mention Word’s grammar checker (and no doubt others) is turning it into reality.

Fowler starts out the relevant section by taking usage writers down a peg or several: “What grammarians say should be has perhaps less influence on what shall be than even the more modest of them realize ; usage evolves itself little disturbed by their likes & dislikes.”


“that” or “which”? Using “which” in restrictive or defining relative clauses (1/∞)

Can you use which in defining (or restrictive) relative clauses?

For example…

“In 1957 work began, under the editorship of R. W. Burchfield, on the new supplement, superseding that of 1933, and treating all the vocabulary which came into use while the main dictionary was being published or after its completion.”
(From The Oxford Companion to English Literature [2000], referring to A Supplement to the Oxford English Dictionary)

In a nutshell, if you’re in the U.S., no (almost certainly); if you’re in the UK, yes, you can, as in the example shown, but many people think you can’t.

“Change nothing in your editing that you do not know to be essential or believe to be beautiful.”

I’ve got a bee in my bonnet. It’s buzzing around in a mad sort of OCD way, and I can’t swat the varmint, try as I might. The apian interloper in my titfer is this: I think changes should only be made to a piece of writing if they are either essential from a strictly grammatical (e.g. verb concord) or meaning point of view, or stylistically desirable. To adapt William Morris’s famous phrase, “Change nothing in your editing that you do not know to be essential or believe to be beautiful.”

I’ll pass over the stylistics here, but one facet of what I mean by “essential” is that truly ambiguous wordings or structures have to be changed. However, such cases are rare; the example with which at the start of this blog is not, to my mind, one of them.

I presume the mother, if American, will confiscate the present, and only give it back when the child replaces “which” with “that”.

Is this change necessary?

I do a lot of editing, and I also review other editors’ edits of articles for academic journals.

One of my oft-repeated comments directed at certain editors is “Is this change necessary?” On a similar tack, I recently copy-edited about half [don’t ask] of a book by a writer and journalist who has already had several books published and writes with flair and distinction. They [Isn’t it handy when “singular they” conceals gender!] then wanted me to review the proofs, which the publisher had had proofread.

Excellently proofread they were, too, complete with a useful, comprehensive list from the proofreader, which explained their decision on style issues such as the treatment of names and titles.

One of the notes, however, read “I have changed a few instances of ‘which’ to ‘that’ were perceived to be a relative clause.” This was a red rag to my bull.  I happened to notice one such change, as follows:

“That the phrase ‘native place’ is still used, however, shows that many Indians are migrants, albeit internal migrants. Such migration, ironically, has been greatly facilitated by the railways which were developed by the British,  a classic example of how they changed India for the good but still made the ‘natives’ feel inferior.”

That word which [my emboldening] had changed to that in the proofs. The book was being published by a British publisher: the change was, therefore, by my lights, totally unnecessary. What is more, it changed the words which/that had come naturally to the author and so, one could argue, changes their “voice”.

(From now on, I will use which/that to highlight restrictive or defining clauses.)

Back to basics: restrictive and non-restrictive relative clauses

Let’s look at the example just mentioned. “Such migration, ironically, has been greatly facilitated by the railways which were developed by the British, …”

Now, which Indian railways are we talking about here? Why, only the ones the British developed before their departure in 1947. The specification is important, because, since then, the Indian government has massively expanded the rail network. So, what the clause “which were developed” is doing is to restrict the extension (in its logical meaning) of “railways”, or to define the kind of railways in question. That is why such clauses are called restrictive clauses or defining clauses.

Now let’s return to the example which/that heads this blog.

“In 1957 work began, under the editorship of R. W. Burchfield, on the new supplement, superseding that of 1933, and treating all the vocabulary which came into use while the main dictionary was being published or after its completion.”

“Vocabulary” here is being restricted to or defined as that which came into use during the long period it took for the original OED to be published (1884–1928), or thereafter. In other words, what is excluded by the defining clause is words that were already in the language before 1884.

How to identify restrictive or defining clauses

One way to identify defining or restrictive clauses which/that is often mentioned is to ask whether removing them changes the meaning of the sentence, or makes it nonsensical. Applying that test to our two example sentences gives:

“Such migration, ironically, has been greatly facilitated by the railways which were developed by the British, a classic example of how they changed India for the good but still made the ‘natives’ feel inferior.”

This is clearly a nonsense, since the subject of “they changed” now becomes the railways.

The other example still makes sense with the clause removed, but the meaning has changed drastically to include all the vocabulary of English.

“In 1957 work began, under the editorship of R. W. Burchfield, on the new supplement, superseding that of 1933, and treating all the vocabulary which came into use while the main dictionary was being published or after its completion.”

So, what are non-restrictive or non-defining clauses?

As the Collins Cobuild Grammar helpfully explains them, they “give further information which is not needed to identify the person, thing, or group you are talking about.”

(Note, incidentally, the use of which in the above restrictive/defining relative clause. The Grammar was produced at Birmingham University, and whoever wrote that section will have been a British English speaker. The use of which was natural to them.)

The Grammar then continues: “If you say ‘I saw Kylie Minogue’, it is clear who you mean. But you might want to add more information … , for example, ‘I saw Kylie Minogue, who was staying at the hotel opposite’. In this sentence, ‘who was staying at the hotel opposite’ is a non-defining relative clause.” Note that the comma here is obligatory to separate such a clause from what precedes.

If you’ve ploughed/plowed through this, you might need cheering up, so I throw in this picture of KM gratis, free and for nothing.

The gold hotpants which/that caused quite a stir when Kylie first exhibited herself in them.



-ise or -ize? (3/3) In praise of monetize, diarize, etc.

-ize verbs are ‘like lavatory fittings, useful in their proper place but not to be multiplied beyond what is necessary for practical purposes.’


Some people have an almost pathological aversion to certain words ending in -ize and would do all they could to expel them from the body of English.

(NB: ‘-ize‘ here stands also for the spelling –ise)

Why? Sometimes it seems almost like a blood feud: just as venomous and visceral, and just as unreasonable.

Here’s an example:

Monetize: a word we didn’t need

Only in the perverted world of the web can something as simple and fundamental as making money be in need of a fancy word like “monetize”

from the blog Signal v. Noise.

Here’s a question from the Grammarphobia blog.

Q: A curator at the Museum of Modern Art in New York was quoted as saying that “risk has been incentivized.” Yuck! Any comments?

A: Someone in the arts has no business using that kind of bureaucratese. Leave it to the CEOs and politicians.

And here’s Brian Garner on disincentivize: “Disincentivize is JARGON for discourage or deter” and he gives the example (from the San Francisco Daily): “We’re competing with Los Angeles and New York firms for talent,” Bochner said. “We don’t want to disincentivize people from coming here because there are huge gaps in salary.”’

Any discussion of words such as the above, it seems to me, has to attempt to answer at least the following questions:

  1. what do they really mean?
  2. are they necessary or useful?
  3. are they overused?
  4. when are they appropriate?
  5. who dislikes them, and why?
  6. when did the dislike start?

A history of contempt

Verbs in -ize have existed in English for a very long time, e.g. baptize since 1297, organize since 1425, generalize also 1425, etc., etc.

The OED lists no fewer than 2,315 of them. Some are nonce words (to wondernize – ‘to make a wonder of’, 1599; to miraculize – ‘to transform [a person] with miracles’, 1751); many were—some might say ‘thankfully’—short-lived (to abastardize, ‘to declare [someone] illegitimate’], 1574—1692|; to accowardize, ‘to render [someone] cowardly’, 1480—1642).

But many are indispensable in everyday language, and seem to ruffle no feathers, e.g. authorize (first recorded in the 14th century), civilize (17th), memorize (16th), sterilize (17th), terrorize (19th), and, more topically, computerize (1960).

One prolific coiner of –izes was the Elizabethan maverick writer Thomas Nashe, whom the OED credits with 28, including overprize, which has survived (by the skin of its teeth), and unmortalize (= ‘to kill’), which has not.

The OED entry for the -ize suffix suggests that he was criticized, nay, anathematized, and martyrized for its overuse:

Reprehenders, that complain of my boystrous compound wordes, and ending my Italionate coyned verbes all in ize.

What happened between then and the nineteenth century I don’t know, but usage gurus in the 1800s repeatedly condemned them, as  Merriam-Webster’s Concise Dictionary of English Usage explains.

Not even Noah Webster himself was immune to izeophobia. While deigning to enter the word jeopardize, he nevertheless noted: ‘This is a modern word, used by respectable writers in America, but synonymous with jeopard and therefore useless’.

If you are enjoying this blog, there’s an easy way for you to find out when I blog again. Just sign up (in the right-hand column, above the Twitter feed, if you’re reading this on laptop, and under the blog if you’re reading it on a tablet, mobile, etc.) and you’ll receive an email to tell you. “Simples!”, as the meerkats say. I blog regularly about issues of English usage, word histories,  writing tips, and Spanish.

And I also edit other people’s work. So if you need some help, don’t hesitate to contact me.


A generalized dislike? ‘Crude, overused, or unnecessary’.

In his 1996 edition of Fowler’s Modern English Usage, Robert Burchfield referred to ‘The widespread current belief that new formations of this kind are crude, overused, or unnecessary’. (He substantiated this by referring to a single comment in Gowers’ Plain English, so perhaps ‘widespread’ should be interpreted as ‘prevalent among the small group of Oxonians, usage pundits and others who care deeply about such things’.)

However, his adjectives reflect some of the issues about these words that I touched on at the beginning.

Are they necessary? To my mind, their very existence confirms their necessariness. Speakers do not generally create phantoms. Moreover, several such words have highly specific technical or scientific meanings. How many people object to being anaesthetized before an operation? (Though of course, the pedantic could insist on being ‘given an anaesthetic’.)

Are they overused? I don’t even know how one would begin to answer this question. If ‘overused’ means ‘there are too many of them’, how many would be not too many? How many would be too few?

Perhaps it means individual verbs are used too often. In which case, there must be a notional cap on any given word. If so, who decides what it is? (‘OK, Mr Carney, you’ve used “undercapitalized” three times today. That’s your lot, mate. You’ll have to find another word, or put a tenner in the -ize box.’)

The question doesn’t make any kind of sense.

Another criticism sometimes levelled at –ize verbs is that they are ugly (e.g. by Gowers), or inelegant. But aesthetic criteria in language are subjective. Your ugliness can be my practicality.

One might be on firmer–though still rather subjective–ground in suggesting that some of them sit best in certain kinds of discourse.

For instance, Garner might have a point that ‘disincentivize’ is jargon and needlessly ousts simpler words such as deter or discourage. Then again, he might be wrong; it all depends on context. In the specific example he quotes, the subject matter is, after all, financial, and you could argue that disincentivize is actually more accurate and focused than the synonyms he suggests: it packages a more complex idea, which means that the sentence could be paraphrased as ‘we don’t want to remove whatever possible incentives we can provide for talent’.

Like many original technicisms (e.g. neurotic, semantic, mesmerize), such words escape the confines of their original domain. That they do so does not make them unnecessary or suspect.

I would, in fact, argue that many -ize verbs are a very convenient way of packaging in one word meanings and connotations that would otherwise take several.

They are beautifully (or uglily, for many) economical.


What do they mean?

One can only take them individually.


Going back to ‘monetize’, many of the 44 comments on the website mentioned at the beginning quibble over what it ‘really’ means1.

Some argue that it is just a pompous way of saying ‘make money out of ’. If so, any flab it adds in pomposity it quickly works off through brevity.

Economy of effort should never be underestimated in language, as a couple of other commenters (?) are quick to grasp. One says: ‘“How can we monetize this?” actually means “How can we make this make money?” and is thus more efficient and avoids the double use of “make”’. Another quips ‘Monetize is a word that has a specific meaning when used in context. It is [a] useful word for making conversations shorter, therefore making meetings shorter.’

But I don’t think it usually means merely ‘make money out of’. As one of the commenters says, ‘the term monetize is more referring to “how can we take this thing we already have (traffic, users, etc.) and convert it into money.”’

That echoes the relevant OED definition and examples: To exploit (a product, service, audience, etc.) so that it generates revenue.

1998   Boston Globe 14 Jan. c6/6   It’s all about eyeballs, audience acquisition… Growth lies in the ability to monetize those eyeballs.

Moreover, that meaning is the fourth and last of a word that first saw the light of day in 1867.

(And if anyone can think of a way of monetizing this blog, do, please, let me know.)



Which leads seemlessly (only joking, but it’s a common enough eggcorn) to another word that is, in my view, both economical and versatile. In 1982, Burchfield described prioritize as

‘a word that at present sits uneasily in the language’. While some people still consider it an uninvited guest, it seems to have made itself at home and got its feet well under the table.

Consider its usefulness. With a single word you can express the meaning ‘Designate or treat (something) as being very or most important’ (e.g. the department has failed to prioritize safety within the oil industry)


‘Determine the order for dealing with (a series of items or tasks) according to their relative importance’ (e.g. ‘age affects the way people prioritize their goals’)


(intransitively) To establish priorities for a set of tasks. (e.g. A hot file forces you to prioritize because you have to select which things will be included.)

Its other benefits include nominalization as prioritization, and derivatives, reprioritize and deprioritize.




Finally, in this paean to -ize verbs, take a word which, as it happens, is more common in British than in American English, despite probably sounding to many Brits like an Americanism; and, far from being new, was first used—albeit in a different meaning—in 1827: diarize/diarise.

It expresses ‘to put in one’s diary’ in a single word. How convenient is that?

Lavatory fittings?


In his The Complete Plain Words (1954), Sir Ernest Gowers, drawing on the well-established ‘unwanted alien’ trope for language, wrote:

‘The main body of the invasion consists of verbs ending in ise.

‘“There seems to be a notion”, says Sir Alan Herbert, “that any British or American subject is entitled to take any noun or adjective, add ise to it, and say, “I have made a new verb. What a good boy am I.”

‘Among those now nosing their way into the language are casualise (employ casual labour), civilianise (replace military staff by civil), diarise (enter in a diary), editorialise (make editorial comments on), finalise (put into final form), hospitalise (send to hospital), publicise (give publicity to), servicise (replace civilians by service-men), cubiclise (equip with cubicles), randomise (shuffle).’

As happens with such verbs, three have disappeared together with their referent (civilianise, servicise, cubiclise), but the others have forcefully demonstrated their usefulness.

Gowers then uses the aesthetic argument:

‘This may be symptomatic of a revolt against the ugliness of ise and still more of isation, which Sir Alan Herbert has compared to lavatory fittings2, useful in their proper place but not to be multiplied beyond what is necessary for practical purposes.’

(He also quipped: ‘If nobody said anything unless he knew what he was talking about, a ghastly hush would descend upon the earth.’)

As long ago as 1996, Burchfield proved that

‘Any feeling that the language is being swamped by new formations in -ization and -ize does not appear to be supported by the facts.’

(As an example, of the 5,219 post-1970 words in the OED, a mere 40 are -ize verbs. )

1 A wag among the commenters writes: ‘The first time I saw that word, I thought “Monet-ize”? You mean, scrunch up your eyes to make everything blurry, like the plein-air painters do? When I learned what the word was intended to mean, I realized my initial thought was correct – it is linguistic bullshit designed to obfuscate the fact that you are trying to figure out how to make money from something that should just be free.’

2 I have to confess, since coming across this phrase, I’ve never understood exactly what Sir Alan meant. Bidets? Toilet paper holders? Bog brush?