There are sev­er­al fac­tors in play here.

Dif­fi­cult con­so­nant clus­ters are of­ten re­duced in rapid speech or over time; think of friend­ship, spend­thrift, twelfth, months.

Much of the dif­fer­ence be­tween an un­voiced and a voiced stop in English is ac­tu­al­ly not its voic­ing but its as­pi­ra­tion, and be­cause one nor­mal­ly on­ly as­pi­rates stops that are both un­voiced and which be­gin a stressed syl­la­ble, you have just lost the prin­ci­pal dis­tin­guish­ing fea­ture.

When you have two con­sec­u­tive stops that dif­fer on­ly in voic­ing, these are es­pe­cial­ly like­ly to fuse, with the first of the pair dropped. Without an au­di­ble re­lease, there is noth­ing to mark the end of one and the be­gin­ning of the next.

Here is a set of words or phras­es where you nor­mal­ly sup­press one of the two ad­ja­cent stops that dif­fer on­ly in voic­ing:

  • cup­board
  • rasp­ber­ry
  • black­guard
  • back­ground
  • post­doc
  • post­dat­ed check
  • sub­poe­na
  • next-door neigh­bor
  • last-ditch ef­fort
  • best dog­sit­ter

It is not al­ways the first of the two that is sup­pressed. For ex­am­ple, no­tice how in back­ground noise, it is the g that ap­pears to get lost: it sounds more like back round.

In con­trast, in black­guard (when pro­nounced as though it were spelled blag­gerd) it is the first of the two ad­ja­cent stops that seems to go away, mak­ing it work like cup­board and rasp­ber­ry with their lost p.

A lost dog may well come out sound­ing like a loss dog in rapid speech, and a black glass like a black lass.

This is not com­plete­ly guar­an­teed, es­pe­cial­ly in new com­pounds whose mor­phemic bound­aries are still clear. It is al­so more apt to hap­pen when the stress is on the first syl­la­ble than when it’s on the sec­ond. But on­ly very care­ful speak­ers will gem­i­nate stops when go­ing out­doors: the t be­comes at most a glot­tal stop — if that. So an out­door the­ater might be said [ˌäʊ̯ʔ.doɻʷ ˈθiː.əɾɚ].

But even a big kite, a bad turn, or a job posting is li­able to lose the first of the paired stops in con­nect­ed speech, since the sec­ond stop is as­pi­rat­ed and the first gets no au­di­ble re­lease.


Just to emphasise the pronunciation guides that people have given elsewhere, it's not pronounced as "cup-board" or "cu-board" but really "cubbered" very similar to "covered".

You have to really think of English as 2 separate languages; the spoken one that has dynamically evolved for a thousand years and the written one which was codified 500 years ago into standard spelling. Over time, the pronunciation is going to drift further and further away from the spelling such that the written version of a word will contain virtually no clue as to how it's pronounced - but will just serve as a generally-accepted "code" that we all know and understand.

It has the added benefit that the "code" will be an endless source of fascination for people like us explore how our "cubbered" must have evolved from a board that cups were put on, that gained some sides, then a top, then, finally, some doors.

In short, in Britain today, there really is no "p" in "cubbered" - except in the archaic spelling "code" that we use to represent it. You may not like it but that's the way it is.

There will always be people at the forefront of the spoken evolution - and those lagging behind. It is interesting that we never seem to hear people campaigning for the proper pronunciation of "knife" as "k-neef" as it "should" be said. Even better, "knight" as "k-nichhter". There are countless, no doubt better, examples if I could think of them.