Haf and have

[Laurence Horn]

At 11:44 AM -0500 3/25/03, Mark A Mandel wrote:
>On Mon, 24 Mar 2003, Duane Campbell wrote:
>
>#It seems that when "have" is used to mean "need" -- as in, "I have to go
>#to the bathroom" -- it is pronounced "haf". But the same speaker using
>#"have" as perfect tense or when meaning "possess" the "v" is vocalized. I
>#have even heard, "I haf to have this."
>#
>#Am I imagining this? Is this one word becoming two distinct words by
>#meaning and pronunciation? Am I just a dilettante observing something
>#that has been studied to death?
>
>I've been aware of this for many years. ISTM that the regressive
>devoicing originated in lexicalized "have to" /'h ae ft@/, as a purely
>phonological phenomenon owing to the loss of word boundary, and is now
>so strongly attached to this combination that it persists even when
>stress and prosody create a break, e.g.,
>         I really HAVE to go to the bathroom
>  with high-falling tone on "have", and the same duration for
>         real    ly
>  and
>         have    to
>  piecewise.
>
But it hasta be lexical/morphological, not purely phonological, given
the voicing in e.g.

There are some books I have to read (but I don't have HAFTA read
them, so I'll go out to a movie instead)
vs.
There are some papers I hafta grade (so I can't go out to a movie)

or
What do you have to eat?   (i.e. have on hand)
vs.
What do you hafta eat?  (given your new diet)

On some accounts, there's a trace in the first member of each of
these pairs, but in any case they involve not the lexical item
"hafta" = 'must' (more or less), but real main verb "have".  I'll bet
Arnold can give us references on where this is all discussed in the
literature.

Larry