Oxford University Press's
Academic Insights for the Thinking World

Kneading bread for the needy

The oddest English spellings, part 18: knock and climb

By Anatoly Liberman

In those rare cases in which people ask my advice about good writing, I tell them not to begin (to not begin?) their works with epigraphs from Mark Twain or Oscar Wilde, for the rest will look like an insipid anticlimax, and, disdainful of ground-to-dust buzzwords and familiar quotations, I also suggest that people avoid (naturally, like the plague) such titles as “A Tale of Two Friendships/ Losses/ Wars,” etc. and resist the temptation to begin those tales with the reminder that it was the best of times and the worst of times.  Despite knock and climb in the title above, what follows is not a tale of two words; nor does it contain a reference to the best (worst) of times in the history of the English language.  The subject matter of the post is a dramatic and largely unexplained simplification of consonant groups in English, and it is a matter of opinion whether that process was beneficial, detrimental, both beneficial and detrimental, or none of the above.  Only one thing is clear: in dozens of words, we write letters that no one needs.

Initial gn- and especially kn- fared badly in all the Germanic languages.  Few words began with gn-; only kn- and hn- were common.  Providentially, no traces of hn- are left in our spelling, but kn- stayed (as always, inconsistently), and, as a result, we struggle with pairs like nave and knave, night and knight.  Lovers of puns may add not and knot, no and know, Neil and kneel, new and knew.  German has also lost gn-, so that nagen is not too different from Engl. gnaw (but no g- is written in German!), whereas Knie and kneten, for example, which look like Engl. knee and knead, are pronounced in accordance with their written image, that is, with kn-. The German examples show that kn- could have survived, and no reasonable explanation exists of why in late Middle English and early Modern English initial groups were destroyed with such severity.  (After a vowel no one minds gn and kn: compare agnail, ignite, acknowledge, sickness, and the rest, but taken became ta’en and was even spelled tane; only later it was restored to its “correct” etymological form with k.)  Words beginning with wr- (write, wreck, and so forth) were the first to get the ax(e), and today none of them are pronounced with w-.  Relatively few people still distinguish between which and witch.  The modern spelling of such words (what, where, while) has it backwards, for at one time they began with hw-, a consonantal group like any other.

The evidence of modern dialects and contemporary documents makes it possible to reconstruct the gradual disappearance of initial groups.  In principle, gn- violates “the spirit of the language,” as they used to say, to such an extent that even gnu (the antelope) and gneiss (the rock) are homophonous in American English with new (or noo-) and nice, but, as always, some speakers used to resist the change.  In a few areas of Scotland, kn- and gn- were attested at the end of the 19th century (nothing may have happened to them since then); in others, kn- became tn- (tnead “knead”).  In the second half of the 16th century, some people wrote knaw for gnaw.  Perhaps around that time  kn- became indistinguishable from gn-, or no consonant at all could be heard before n, and the writer tried to guess which of the two would be more suitable (naturally, both would have been wrong).   By the close of the 17th century most English speakers had shed k and g before n, but even a hundred years later foreigners recorded tnocked and tnots for knocked and knots.

It is curious to observe variation and occasional reverse spellings in some place names.  Alongside Knotting Bernes (1476; Middlesex), one finds Notingbarnes (1517; now it is called Notting Hill), but Knokholt in Kent was once Nokholt, because its n- is the product of redistribution (as in nuncle for mine uncle and nickname for an eke-name): n- goes back to then, while ok- is the continuation of the word for “oak” (Middle English at then ok-holte “at the oak grove”).  Engl. Holt, contrary to German Holz and Dutch hout, is now remembered mainly as a family name.  If you know a Mr. / Ms. Noakes or Akehurst, inform them that their ancestors lived in the vicinity of oaks (they probably know it).  When people begin to write knot for not, we may conclude that k before n is either disappearing or has disappeared, but a vague memory of it is still alive.  Punning is a natural consequence of mergers.  I once lived near a place called Naughty Pines.  The pines around were bent and knotty.  By the way, Nottingham began its life as Snot’s home, Snot being the name of the place’s owner.  Its s- is gone.

Unlike gn- and kn-, initial gl- and kl- have changed little, if at all.  Yet the widespread dialectal variant tlos “clothes” shows that kl- was also caught by the torrent, and Noah Webster (1828) gave more examples like it from American English.  One could expect that final consonants would be treated even less mercifully than initials ones, for the general tendency was to chop everything that followed the root.  However, this is not what happened.  Here the most noticeable event was the simplification of -mb.  I have often written about the history of numb, dumb, and thumb and will mention such words here only because this change should be seen as part of a larger whole.  The destruction of -mb was total: tomb, lamb, climb, and even jamb, from French jambe.  Final -mb became as impossible as initial kn-.  The earliest spellings of comb as come and coame occurred in the 15th century.  Combing has no b in pronunciation; other than that, in the middle of the word mb is allowed (chamber, embers, rhombus, and compare Lambeth, a place in Surrey: its Old English form meant “lambs’ pier”). The history of final -ln, -nd, and -ld is less exciting (note only that lawn “a piece of ground kept mown” was at one time lawnd).  Only final -ng underwent a dramatic change, but this is a special plot.

In discussion of etymology we often end up saying “No one knows where such and such a word came from.”  Here all the facts have been collected, but “no one knows” why the events described above happened.  To be sure, people are lazy and try to pronounce less and less, but the factor of laziness is permanent, while the simplification of consonantal groups occurred several centuries ago.  One is also tempted to say that kn-, gn-, and -mb were inconvenient to pronounce.  This is true, but were they convenient in the days of Chaucer?  Our organs of speech have not changed since 1400, the year of his death.  Non-specialists will now understand why linguistic manuals are so fond of the statement: “The causes of sound change are unknown.”  Not everything is so bad, but it is bad enough to cause wonderment and irritation.

Other odd English spellings…

Anatoly Liberman is the author of Word Origins…And How We Know Them as well as An Analytic Dictionary of English Etymology: An Introduction. His column on word origins, The Oxford Etymologist, appears here, each Wednesday. Send your etymology question to him care of blog@oup.com; he’ll do his best to avoid responding with “origin unknown.”

View more about this book on the

Recent Comments

  1. John Cowan

    Combing has no /b/ because -ing is a uniformly applied inflectional suffix rather than a derivational one; it is normally only derivational suffixes which preserve such irregularities. The exceptions are: a few weak verbs with irregular shortening in the preterite, namely kept, slept, crept, wept, dealt, felt, knelt, meant, bled, bred, fed, red, sped, led, met, slid, lit, shot, left, lost, fled, said, shod and their derivatives; and the nouns with f/v alternation, namely calf, dwarf (in some senses), elf, self, shelf, hoof, knife, life, wife, leaf, sheaf, loaf, scarf, thief, wharf, wolf and their derivatives.

  2. mollymooly

    @John Cowan: there is also “limn”, where some people sound the n in “limning”.

Leave a Comment

Your email address will not be published. Required fields are marked *