Part 1: Vowel and Consonant features
seems fine. there's the question of what to do with syllabics, /=l =m =n
=r/. I would prefer to treat these as sequences of /@/+ consonant, ie.
/=l/ = /@l/; syllabicity + laterality (etc.) is a product of temporal
phonetic interpretation, not a claim about the phonological status of
syllabic consonants.
Part 2: Legal Onsets and Codas
(Could we please call them codas not offsets?)
The onsets allow for Cj- as clusters, presumably analysing "beauty"-type
words as bj+u rather than b+ju, where /ju/ is a diphthong. I don't know
whether it matters much in the long run, but I'd prefer it if the
phonology were /C + ju/ rather than /C + j + u/. I seem to remember that
when it comes to temporal compression /ju/ is not like the others. (Not
sure what that means -- perhaps it's a reason not to have /ju/ as a
diphthong?!)
The codas contain material which properly speaking belongs in an appendix.
See page 10 of my "Syllable Structures" document; we want to treat
appendices as a sub-type of coda, with only (cnsgrv = n cnscmp n) as a
possible place of articulation in the Appx. The clusters that need
removing from that list are: /bz dz gz mz nz Nz fs vz Ts Dz lz mps nts Nks
mbz ndz Ngz lps lts lks lbz ldz lgz mT mpT mps mfs mTs mpTs ntT nts nTs
ntTs nTs nkTs/ and perhaps /NT NkT/ although 'strength' and 'adjunct'
provide evidence in favour of these as codas proper.
By the way, if all those clusters with /s z/ are included, then I think
similar clusters with /t d/ should also be included. But in the
appendix-analysis we agreed on, /s z t d T/ all go in the appendix when
the word is morphologically complex: we want to be able to distinguish,
phonologically, between "tax" and "tacks", and the Appx. allows this.
Part 3: material for a lexical entry.
I'm not sure I can see what this is about: is it constraints on feature
combinations? why are all the values set at 'n'? (is it just for
illustration?)
Part 4: Syllable parses
These look fine. One question though: why are the syllable-final
consonants marked as ambi = 'y' ? Is this because they are *potentially*
ambisyllabic? I'm not sure it makes any difference for utterance-finals,
but this might be consequential when we put words together, where we want
to do the things outlined on page 8 of the "Syllable Structures" document.
Thanks Mark for doing this parsing biz. Let me know if there's more stuff
I can help you out with.
Richard
Richard Ogden
rao1@york.ac.uk
http://www.york.ac.uk/~rao1/
On Tue, 3 Mar 1998, Mark Huckvale wrote:
> Attached are four components of the lexical parser. This
> is the program that I am developing to take the SAMPA transcription
> in the lexicon and expand them to syllable structure.
>
> Part1: listing of vowel and consonant features for SAMPA transcription
> Part2: list of legal onsets and offsets
> Part3: XML DTD for a lexicon entry
> Part4: partially completed lexical entries for two words
>
> I have yet to:
> - put features on <SYL>, <ONSET>, <RHYME>, <NUC> and <CODA> tags
> based on the segments beneath them
> - deal with syllabic /l/
> - deal with linking /r/
> - make agreements of CNSVOC etc within onset/coda
>
> But nevertheless, comments on what we have so far welcome!
>
> Mark
>