to hyphenate or not to hyphenate?

The Shorter Oxford Dictionary (6th edn) recently made the news for deleting a lot of hyphens that had been in the previous edition. According to the AskOxford website:
Drawing on the evidence of the Oxford Reading Programme and our two–billion–word Oxford English Corpus, we removed something like 16,000 hyphens from the text of the Shorter. So it's double bass, not double–bass, ice cream not ice–cream, makeover instead of make–over, and postmodern rather than post–modern. [Italics added because it was driving me crazy that Oxford hadn't marked the self-referential use of these words!]
Now, I neither have the two editions of the Shorter Oxford, nor would I have the time to look up all of the de-hyphenated words if I did have them, but it's long been my impression that British dictionaries (and possibly BrE speakers--we'll come back to this below) and American dictionaries (and speakers?) differ in their relationships with hyphens. When the Association of British Scrabble Players switched over to the international dictionary (including the former US and UK Scrabble dictionaries), one thing that struck a lot of players was how many more verbs could take the re- prefix. (As in relocate or reassemble. I'm afraid I don't still have my old dictionary to tell you which ones weren't allowed.) This was in part because of the BrE tendency to put hyphens between the prefix and the base verb, especially in cases in which not to do so would involve the same letter repeated twice at the end of the prefix and the beginning of the base word. So, BrE prefers re-elect, which is happy without a hyphen in AmE: reelect. The same thing happens with the prefix co-, especially before another o, so that BrE tends to prefer co-ordinate and co-operate, whereas AmE prefers coordinate and cooperate. So, I wondered, do the changes in the Shorter Oxford reflect more AmE-like use of hyphens? I found the following examples of de-hyphenated words in the Shorter Oxford 6 from news items and commentaries about the change: this BBC article, this New York Times article, World Wide Words and the aforelinked OUP press release. Then I compared them to the American Heritage Dictionary, which happens to be on my desk.

Shorter Oxford 5 (2002)Shorter Oxford 6 (2007)American Heritage 4 (2000)
fig-leaffig leaffig leaf
pot-bellypot bellypotbelly
double-bassdouble bassdouble bass
ice-creamice creamice cream
hobby-horsehobby horsehobbyhorse
fire-drillfire drillfire drill
water-bedwater bedwaterbed
test-tubetest tubetest tube

The bold entries in the table show the three cases in which the change in SOED6 is a change in the opposite direction from the AHD4 entry. (And I have to take issue with the AHD's one-word status for hobby horse. Not how I would spell it. I'm less-than-sure about potbelly too.)

Does this mean that BrE is becoming more like AmE?
These changes probably have at least as much to do with the SOED looking more carefully at how these words appear in printed language as they do with any actual language change. After all, there have been only five years between the editions--that's an awful lot of hyphens to bite the dust in such a short time. The NYT article notes, "That ice cream and bumblebee ever had hyphens to begin with suggests an excess of fussiness on the part of older lexicographers" and BrE-speaker Michael Quinion at World-Wide Words says, "The new SOED lists many hyphenless words such as leapfrog, bumblebee, crybaby, pigeonhole, lowlife, and upmarket, which will be a relief to those of us who have been spelling them like that all along."

I do get the feeling, however, (and this is just a feeling) that BrE favo(u)rs keeping words more separate. We can describe a hierarchy of 'one-wordiness' or 'joined-up-ed-ness' of English compound nouns, like this:
fully integrated: lifeboat, prejudge
semi-integrated: sit-in, semi-integrated
not integrated: ice cream, throw up
Using this hierarchy, I'd suppose that BrE writing tends toward(s) non-integration--that is to say, keeping words separate, or at least hyphenated, whereas AmE is happier to have more fully integrated compounds. It's just a hypothesis, though, and you're welcome to test it. (Hey, final-year students! There's a project!) Part of the reason I've formed this hypothesis is the widespread habit in BrE writing of treating some prefixes as separate words. Here are some examples, in which in AmE I'd have to have at least a hyphen, if not a single word, but which one sees not infrequently as separate words in BrE:
over- as a verb prefix
On the other hand, children are very good at expressing what motivates them in a learning context but perhaps over egg the custard a little when it comes to saying that what they find boring. --Scaife & Rogers, 'Kids as Informants'

post- as a prefix meaning 'after'
Public Health Advice ? Post Flood
Before re-occupying your home
The flood water affecting your home or other property may have been contaminated with sewage and other contaminants... --Bridgnorth District Council

as a prefix meaning 'below'
Sub normal growth rate (usually a height velocity below the 25th centile usually. equates to less than 5 cm per year in a pre pubertal child). --from an NHS [Wales] diagnostic guide
And so forth. This is the kind of thing I find myself 'correcting' constantly in student work in the UK (never as much of a problem in the US), so much so that I started to wonder whether I was the one in the wrong in my new dialect-land (as has happened before). But no, my New Oxford Dictionary of English treats all of these as prefixes, requiring hyphens or full integration with the base word. Fowler's Modern English Usage (3rd edn) doesn't mention the writing of prefixes as separate words, and in its entry on hyphens recommends the use of hyphens with prefixes--as opposed to full integration--in the cases where the prefix joins to a proper name (anti-Darwin) or where the same letter is repeated (re-elect) or an ambiguity ensues (as in the pro-verb/proverb case that I mentioned a while ago).

But before you go any further in thinking about this, I give you the following cautionary quotation, cited on the American Dialect Society e-list recently:
If you take hyphens seriously you will surely go mad.
--John Benbow, Manuscript and Proof, 1937


  1. Personally, I think hyphens are like antique overpunctuation. Fussy and unnecessary for clarity, soon to be obsolete, I hope. I avoid them mercilessly.

    1. So restore or re-store - to return to its original state, or to store again. Remove the hyphen and it isn't as obvious which is the intention until you read more context.

    2. I am seeing a lot of odd hyphenation in older British writings; I just came across "ar-ticle". Any explanation for these?

    3. That’s the hyphenation break in the Merrimack-Webster dictionary, so there’s nothing particularly British about it. That’s where the syllable break is.

    4. *Merriam. Damn you, autocorrect!

    5. How would not using a hyphen work with the following? I re-sent (the message, email etc) and the case of not using a hyphen - 'I resent (the message, email etc). By not using a hyphen it has a completely different meaning. Surely the prefix is exactly that - a prefix to a word and not the creation of a new word (especially when doing so just duplicates the spelling of an existing word that has it own definition which, in turn, can create confusion).

  2. I'm always thrown by the American word coworker, which I invariably read as cow-orker, before doing a mental double take. As a Brit I would always write 'co-worker'.

  3. But, in my experience in AmE, there are examples of hypenated compund adjectives, e.g. "I like to go first class, so I bought a first-class ticket." But I don't know why.

  4. And the New Yorker uses cöoperate (or possibly coöperate) which really strikes me as weird.

  5. My policy is to write the word or words in whichever way seems least likely to require the reader to read the sentence twice - insofar as I can guess what the reader is used to. My guess is that I'd therefore use pigeonhole, test tube and fire-drill. With both "fire" and "drill" capable of being a noun or a verb, a hyphen seems to me likely to remove doubt the moment the eye alights on it.

  6. Hypens seem in most cases to be overpunctuation, but I can see in cases like coworker how the lack of a hyphen might be confusing. I also have a natural inclination (being AmE) to write it co-worker, though that may just be an aberration.

    Also, shame on the Oxford website for use/mention confusion. Especially as they did it fine earlier on the page. You would think they'd catch that.

  7. I have a book titled "REELECTION OF LINCOLN" (all-caps intentional). The cover art uses a delicate serif font for the title. I was at least halfway through before I realized it wasn't "Reflection of Lincoln" (which explains why the narrative starts in 1863, a fact that confused me immensely). Needless to say, I prefer a hyphen in "re-election" (I speak AmE).

  8. The New Yorker uses coöperate (and coördinate), with the dieresis on the stressed syllable, which I've always found a graceful little habit, referring to the bygone days when they were compound words.

    But I read so much material printed before, oh, 1930 that my instincts as far as punctuation goes are hopelessly old-fashioned.

  9. For all those of you who don't like hyphens, how would you interpret this headline?

    Senator approves anti child abuse program

  10. There's nothing illegal, or even shocking, about cow orking.

  11. John Cowan said:
    "There's nothing illegal, or even shocking, about cow orking."

    Try justifying that to the cows! (or the orks)

  12. BrE (ScE). Thank goodness! I am not alone!! I thought I was the only person in the world who did the cow orker thing. It is such an irrational relief to find I am not.

    Re Interface's headline, I can see what you are getting at (I think), but really I think the headline's meaning is crystal clear, as the chances of a US senator approving anything to do with anti-child abuse are absolutely zero. Context makes the meaning clear here. Unless, of course, you are going to stun me and tell me of a case where a senator DID approve some sort of program(me) for abusing children.

    Lynne, should your "different... than" in the original post not have been flagged as AmE?

    1. Oh my, how times have changed when it comes to US senators condoning child abuse among prospective immigrants.

      Speaking of purported "submersible" and "abusable" entities, the context argument holds a lot of water when it comes to hyphen use in headlines. That being said: the "anti-child-abuse" modifier occurs frequently, even if inconsistently, across single news sites that are widely read and generally respected. A pinch of salt in strictness likely helps in our usage issue, not with the other.

  13. Oh, of course it should have, Cameron. But by the time you look again, it won't be there!

  14. Cooperate (or naive, or any other English word) with diacritics drives me nuts. It doesn't serve any purpose. Maybe there's a word somewhere that needs to be distinguished from another pronunciation, but I can't think of one.

    As for me, thanks partly to the internalistion of my publication's style guide, I hardly ever hyphenate. It boils down to when I can't quite convince myself that it's a single word, but the prefix doesn't stand on its own. So "long term" but "non-payment", although I wouldn't really object to nonpayment.

  15. Aha, here's a good example from the Sunday Times, and it doesn't even involve a string of nouns.
    "Eventually, Watson did Wilson a great service by forcing him to rethink higher level biology, make it less stamp collecting and more of a theory driven science based on natural selection." I read this as far as 'theory' and it makes sense:-
    ".... make it less stamp collecting and more of a theory" BUT THAT'S NOT WHAT SHE MEANT.
    She meant "make it less stamp collecting and more of a theory-driven science". That seems to me undeniably superior, since it doesn't impose on me the change-of-gear and then re-read maneouvre. By contrast, I'm not sure whether a hyphen in stamp-collecting helps. Nor in 'rethink'.

  16. Ugh. That is one ugly bit of 'descriptive' writing anyway.

    As you may well have noticed my punctuation skills are probably a bit basic, grammar wasn't high on the agenda when I was at school. However, I still appreciate the 'nudges in the right direction' that punctuation can give me.

    For example, I would have preferred

    ... make it less 'stamp collecting' and more of a theory-driven science ...

    to help me understand what the writer was getting at :)

    As for cooperate - I always start of reading it as coop erate :) and putting those little dotty things over the ö means nothing to me at all.

  17. So, how about a man who runs a small business; is he a small businessman, a small business man, or a small-business man?

    And I am pleased other people struggle with cow orking too. Another one I have a problem with is noone for no one or no-one. I keep thinking it is a person who spells their name with lower case, like k. d. lang or e. e, cummings.

  18. My formative exposure to BrE writing was largely from the turn of the century to the 1920s, so it seems odd that the New Yorker would be the one using a diaeresis (I still think of 'role' as missing a circumflex, and 'phone' to be preceded by an apostrophe).

    My personal bias is to make a single, unhyphenated word. I was recently surprised to find that 'trainwreck' and 'birdbath' weren't considered to be single-words.

  19. I use "no one" fairly often in speech, and remember when I discovered that there wasn't really a way to write it that wouldn't get me in trouble in my (American) elementary school... I still mostly avoid writing it out of habit, but it makes me a little sad.

  20. According to Wikipedia (so it might not be true) people of Germanic heritage make up the largest single ethnic group in the US. In German and Dutch (the primary Germanic languages) there is a strong tendency to join words together, creating 'compound nouns'. Could this be a factor, albeit subtle, in the trend?

  21. The solution to 'cow orker' is to use "colleague".

  22. I mentioned this in my blog a few weeks ago and didnt know about the differences across different versions of dictionaries. I might have ot ammend that entry.

    Very interesting, as always, though.

  23. Thanks for mentioning 'no one.' E.E. Cummings notwithstanding, I don't generally have a problem with wanting to make this one word, but I am almost continually surprised that 'a lot' isn't one word.

  24. Dearieme:

    I find "colleague" to be academic or highfalutin in tone, and quite unsuitable as a substitute for "coworker". I don't know if this is AmE-specific or not.

    The hyphen that's inserted between the separate words of a compound used as a modifier, as in "all-weather roads" (but omitted in "roads suitable for all weather"), is a different thing from the permanent use of hyphens in certain compounds. Those are essentially a matter of spelling, and follow the same feedback loop as spelling does: dictionaries write down the usage of publishers, and publishers look in dictionaries when choosing forms. Eventually the loop converges.

  25. [BrE] What seem to happen is that when a brand new (brand-new? brandnew?) construction appears it is hyphenated. But when the construction becomes accepted the hyphen is dropped. One recent example would be "e-mail", which is now more usually written "email".

  26. Ah, John, the loop converges - that sounds like a blackhole.

  27. I also note the difference between AmE 'percent,' and BrE 'per cent.' for no other reason than I forgot to last time.

  28. In Mighty Fine Words and Smashing Expressions, Orin Hargraves has this to say about compound nouns being written either as two separate words ("open"), one word ("closed/solid"), or hyphenated: "American English generally is far quicker and more ready to adopt solid compounds, and to eliminate hyphens, than is British English ... A general pattern prevails in which American English is more ready to go from open to closed once usage has been established, foregoing the hyphen stage altogether; and for British English to go immediately to a hyphenated form and never depart from it. But this is not a fast rule..." (pages 30-31, OUP 2003)

  29. Very useful quotation, Nat. Thanks!

  30. Seen today: "multi-wifing".

  31. dearieme - my wife would kill me if I tried that!

  32. Andy- According to the 2000 Census ( in the US, German IS the #1 language. There are three different charts, one of primary ancestry, one of secondary ancestry, and a third of all ancestries, and German is first on all three, though Irish is a close 2nd on the secondary list. Numbers from the 3rd list:
    German: 42,885K
    Irish: 30,528K
    English: 24,515K
    United States of America: 20,625K

  33. Life, this vale of tears, has but few pleasures or compensations. As soon as we are old enough to appreciate the joys of childhood, then we adolesce and crave the intoxicating thrills of growing up. Our majority attained, we at first exult in shouldering the yoke of adulthood, but, sooner or later, that burden starts to weigh heavy, and we stoop and strain. Turning to look over our now-bent backs, we glimpse, receding, the days of our youth. Eye, heart and mind strain to recapture the heady freshness of being young again, but, to our chagrin, we find we can only live it vicariously through the clear looks, translucent skins, shining hair and reverberating laughter of our sons and daughters, our pupils and students, and the few fair ones we may glimpse at twilight – they not even noticing us – from some upper window as we go to get a stamp for a letter or a pill for a pain.

    What does endure, however, is the sheer, stonking whoosh you get from putting a hyphen between two nouns and making a compound from them. Clarkson and his Hamster can keep their noisy, gas-guzzling roadsters. Give me an adjectival noun coupled firmly to its yielding nominal partner any day. Give me traffic-police, can-openers, family-ties, queen-Elizabeth in Buckingham-palace, cream-teas, nylon-stockings, a Danube-steamship-company-captain, moon-landings, harmony-grits, atom-bombs, ring-fingers, fish-fingers, carol-singers, , Brighton-pier, chip-shops, candy-floss, the channel-tunnel, Channel-Four, the British Broadcasting-Corporation, noun-phrases, phrase-structure-rules, language-separation, blog-posts, compound-noun-hyphenation-bores and forum-bans.

  34. "co-operate", "re-elect", and "no one" strike me as the most correct forms. The "umlaut" serves to notify the reader that two like adjoining letters are not a long vowel but straddle a syllable break. Now for my pet peeve: every day vs. everyday. The latter is a one-word adjective (an "everyday" word), but the former (meaning "each day") is constantly butchered and spelled as one word. Aargh!

  35. I came down here to the comments to agree with proponents of the diaeresis.

  36. BrE, Scot, mid 60s. Re cow orking. I have never actually seen the spelling mis-led. But the first few times I saw the word misled, I wondered how you would misle someone. To my young ears, it sounded similar enough to hustle to be plausible.

  37. What led me to this site is my absolute distaste of the over used hyphen. When I am typing in Word and using an adjective before a noun, the spell check joins the words together with a hyphen. Drives me crazy. What do you have against adjectives, I say? Hyphens with prefixes are a different because they are there to aid pronunciation. It's just too easy to think that a 'coworker' is someone employed in the bovine industry. Reelect too easy to miss the second E. But please leave the adjectives alone. And speaking of German and its compound words, how easy is it to pronounce something 7 syllables long?


The book!

View by topic



AmE = American English
BrE = British English
OED = Oxford English Dictionary (online)