> Dear WG list,
> I've just put a new paper on my web site: "Wanna revisited". I give a WG
> account of the old facts about wanna (as in "I wanna go home", = "I want
to
> go home."), plus one tiny little new fact. The main innovation in terms of
> WG is that I give arguments for distinguishing words from forms as well as
> from phonological structures. This is something that And Rosta has been
> arguing for over the last decade or so. Any comments will be
> welcome. You'll find it at www.phon.ucl.ac.uk/home/dick/wanna.htm.
> Dick
The symbol used for the lax high back round vowel [U] looks like the
symbol for a labiodental approximant, rather than the nowadays more
standard inverted small capital omega.
Anyway, I definitely think the paper gets way nearer the mark than
other analyses (e.g. Pullum 1997's weirdness). But there is a key
difference -- in either content or explicitness -- between this
paper's model of the syntax/morphology/phonology interface and the
one I espouse. The difference involves forms. In my model, forms
work thus:
Syntactic words have a stem, which is a phonological pattern.
Syntactic words have a form, which is a phonological pattern.
By default, the form is the stem, but inflection makes forms
differ from stems.
Syntactic words have an enunciation, which is a phonological 'event'.
Enunciations have a blueprint, which is a phonological pattern.
(Enunciation isa blueprint.)
By default the blueprint of the enunciation of W is the form
of W.
If two words share the same enunciation, then by default its
blueprint is a concatenation of the two words' forms.
To the extent that I understand it, the model outlined in the
paper is less explicit than this, so I can't say exactly how
they differ, but the paper does say "wanna is possible (given
the syntactic constraints mentioned above) if, and only if, the
morph {want} is next to {to}". But what exactly does that mean?
It is nonsensical to say that the form {want} precedes the form
{to} *in the sentence*; the phonology plainly goes against it. In
terms of tokens, we have a phonological word that isa the form
/wanna/. The word WANT precedes the word TO, but syntactic words
don't consist of sounds. The solution I'm forced into is to say
that when the syntactic conditions are satisfied, WANT and TO
share the same enunciation, whose blueprint is by default a
concatenation of the forms of the instances of WANT and TO.
Further rules define special alternative forms with reduction of
/tt/ to /t/ or deletion of /tt/. So /wants to/ and /wanted to/
are also single phonological words.
--And.
|