It has been commonplace, since at least Borowsky (1986: ch. 2) and possibly much earlier, that the English word bomb ends in a /b/ which is deleted word-finally, perhaps taking a free ride on the same rule that deletes the /ɡ/ in words that end in a velar nasal (e.g., sing). This /b/, however, is present in two related words: bombard [bəmˈbɑɹd] and bombardier.

There is some distributional evidence for such an analysis because word-final [mb] is quite rare. For what it’s worth, I find final [mb] unobjectionable in the late-learned iamb; CELEX2, which transcribes RP, also gives this pronunciation. But there is no reason to assume that it is bomb-ard in the first place: there are no other words in –ard or –ardier in all of English. Two other possibilities suggest themselves: first, the suffix could itself be /b/-initial -bard, or it might be the case that speakers do not decompose these words at this juncture in the first place. Certainly, I know of no other “English suffix” which selects for exactly one root. Thus the only evidence that bomb ends in /b/ is purely distributional, and reasonable phonologists can disagree how compelling that evidence is. I for one am unconvinced.

References

Borowsky, T. 1986. Topics in the lexical phonology of English. Doctoral dissertation, University of Massachusetts Amherst.