Re: labelling

Paul Carter (pgc104@york.ac.uk)
Tue, 15 Sep 1998 16:04:42 +0100 (BST)

On Tue, 15 Sep 1998, Paul Carter wrote:

> On Tue, 8 Sep 1998, Sarah Hawkins wrote:
>
> > We think the label should be D_0.
> > There would be no D alone in these cases.
> > This solution loses the neat hierarchical organisation of having the "end
> > of the phone" marked by a single letter. The criterion for the "end" will
> > have to be
> > (i) the RIGHTMOST label for the phone
> > except that where there's more than one possible ending, it will
> > be
> > (ii) the rightmost label for the phone that does not contain _#
> >
> > Note: I (Sarah) *think/hope* that this convention has an added advantage:
> > you can of course get voiced and then voiceless /z/, for example. On
> > current criteria, these should be marked z ...... z_0. (not z_v ......
> > z_0 , because z implies voiced). If the end of the phone is the rightmost
> > label (with proviso (ii) above), then using z ..... z_0 is not a problem.
> > (It is a problem if z means end of phone, as well as voiced.)
>
> Sorry not to have been on the ball earlier. It probably isn't a big
> problem, but lab_# subsumes lab_0, so z_0 would fail on count (ii) above.
> Presumably any search mechanisms we may use would just have to know that
> # > 0.

You will of course have noticed that my message is complete gibberish.

With apologies for my brain.

Paul