[fragments of a discussion of righteous outrage regarding
  the use of something rumored to be spelled had've]

>>>What makes this usage especially fatuous, to my mind, is that you can't
>>>even spell it (had've?....had of?) ....

>>Because you can't spell it, it's worse?  Speaking came before writing,
>>you know.

>Considerably worse. If it were just a solecism that could rationally be
>spelled, then its use would merely indicate ignorance of the correct usage
>(for example, "would've" when the subjunctive "had" should be used). But
>"had've" - presumably short for "had have" or "had of" - is an error that
>can't make any sense at all when reduced to writing, which only highlights
>its total absurdity.

They're frequently contractions of phrases that have become so ossified that they're effectively single words, some nonce forms like whatchamacallit, others hyphenation flamebait like flame-bait, some too naughty to use in ordinary writing, like /jÍz@m/; and a select few that are actual ongoing changes in the grammar of English, like woulda, shoulda, gotta, couldna, wanna, gonna, hafta, usta and the like.

There are some new suffixes and prefixes getting ready to be grammaticalized, and in (what used to be) the normal course of events these would generate some new inflectional morphology for English to replace some of what it lost over the last millenium. Technological society, however, is a totally new environment for language propagation and change, and all bets are off.

One of these new suffixes is apparently what gets spelled variously as -'ve, of, or -a, as in should've, should of, shoulda, representing a pronunciation that loses the /v/ almost all the time in rapid speech (listen to what others actually say for a while and then you'll be ready for the shock of listening to what you say instead of what you think you (ought to) say), coming out as a shwa suffix /-@/.

This seems to be tied up in many people's minds with the subjunctive, which is reasonable because it appears most often with modal auxiliary verbs, the lexical instantiation of modality. I've been puzzling over had've since the first post, but in fact I can't remember encountering it in the affirmative. I have encountered hadn't've or hadna in speech, but only in the negative; I think it's a polarity item.

I suspect much of the rancor that greets spellings of had've is compounded by the fact that it's extraordinarily strange in an affirmative context. In a negative context, it's a mimic or variation of the more literate if you hadn't insulted him produced by adding an epenthetic shwa (i.e, one little uh - had-n-a instead of had-n), possibly just to link with the following past participle, and it sounds fairly normal because it sets up a parallelism between the clauses.

    /haedn.@/   ||   /wUdn.@/
    hadn't'a     ||    wouldn't'a

This is a Perfective construction; that is, it uses -- or overuses -- the perfective auxiliary have. Like the perfect, the English subjunctive is ripe for such a restructuring. We have occasions where we'd like to use it, like speakers of any language, but what's left of the English subjunctive morphology is close enough to useless that many people don't bother with it, or misunderstand it when they hear it used, or read it. Certainly they're not taught about it in school. So they go with what they know.

What other choice do they have? After all, they're alive and so is the language they're speaking. And this is not a mere metaphor. There's no such thing as a mere metaphor.

It's not entirely clear whether we speak a language or it speaks us.