Several times recently I have noticed the newspapers referring to ‘slebs’, by which they mean ‘celebrities’. Duly on guard against the recency illusion that leads us to think things we’ve just noticed must therefore be new phenomena, I checked in the OED. I find that the first citation there is from 1996, so a good fifteen years ago.
Shortening a word in colloquial speech is nothing new. Compare bus, phone, mic, etc., and also street cred (credibility) and now peep(s) (people). But what I want to discuss here is the loss of the schwa from səˈleb(rəti) (= DJ’s sɪˈlebrɪtɪ): the word is shortened not to its bare stressed syllable leb, but to sleb.
Most cases of compression involving schwa loss are found in the phonetic environment of a following liquid plus a WEAK vowel, as in historically hɪˈstɒrɪk(ə)li, camera ˈkæm(ə)rə, factory ˈfækt(ə)ri. Hence we regularly find compression in the adjectives moderate ˈmɒd(ə)rət and separate ˈsep(ə)rət, with their weak-vowelled suffix, but not in the related verbs to moderate ˈmɒdəreɪt and to separate ˈsepəreɪt, where the phonetic environment is a following STRONG vowel.
So what we have in ‘sleb’ is not mainstream compression, because the vowel in -ˈleb- is strong. Comparable examples that spring to mind are the colloquial possible loss of schwa in terrific təˈrɪfɪk ~ ˈtrɪfɪk, colossal kəˈlɒsl̩ ~ ˈklɒsl̩, correct kəˈrekt ~ krekt, and perhaps pəˈhæps ~ præps (OED p’raps dated 1745). I can’t recall seeing any discussion of this in writing anywhere, though it’s something I’ve talked about in practical phonetics classes often enough. (There's probably something in Gillian Brown or Linda Shockey’s books on the phonetics of colloquial English.) In the rough-and-tumble of rapid conversational speech I suspect that this reduction can be found for any word with the initial string obstruent—schwa—liquid. But it is presumably much rarer in words such as career, collide, forensic, giraffe, Goliath, Jurassic, Korean, peruse, salacious than in the everyday words mentioned in the previous paragraph.
Very occasionally the reduction becomes lexicalized, as for those speakers whose citation form for police is pliːs rather than pəˈliːs (or dialectal ˈpoʊliːs etc.). There’s also pram, from perambulator, for which the OED’s first citation is dated 1884. Usually, though, we remain aware of the difference in pronunciation in pairs such as plight – polite, crowed – corrode, Clyde – collide, even if we sometimes pronounce them identically.