Re: quantity triggered vs. quantity sensitive stress

A belated reply, but I think not repeating what was in other messages
in the follow-up thread,

Kristian:
> I was just wondering if there are natlangs that, instead of having
> stress that is sensitive to the weight of certain syllables, have
> stress that consistently creates heavy syllables. This appears to be
> a feature of Lumanesian (my conlang) which I have recently
> 'discovered'. Fearing that this is a contrived feature as a result
> of conlanging, I now consult the experts in CONLANG-L. I still would
> like Lumanesian to appear natural. I'll illustrate what I've
> discovered below.
>
> Firstly, I know there are stress systems that function in a manner
> such that heavy syllables nearest one end of a word recieve stress.
> And if there are no heavy syllables, stress is then placed on a
> certain syllable near one end of the word. This is what I mean by
> quantity sensitive. The result is that stress can occur *almost*
> anywhere in a word provided the right conditions are there to
> support it.
>
> For instance, according to "Phonology" by Francis Katamba Ngiyamba
> stress rule dictates that stress falls on the first heavy syllable
> of a word but by default places stress on the first syllable if
> there are no heavy syllables. Heavy syllables in Ngiyamba are
> syllables with long vowels. Thus, one could have stress almost
> anywhere in a word depending where the long vowels are located.:
>
> initial stress: girlambidi "big star"
> antepenultimate stress: "gabadaabidi "big moon"
> penultimate stress: gabadaaga "on moon"
> or final stress: gabadaa "moon"
>
> The situation is quite different in Lumanesian. It appears that
> stress is what triggers a heavy syllable and not the other way
> around. The reason for this is that, unlike quantity sensitive
> stress languages, stress in Lumanesian lexemes is consistently
> penultimate (with the exception of monosyllabic lexemes).
> Furthermore, this stressed syllable is consistently a heavy CVC
> syllable (whereas light syllables are CV only). Stressed CVC
> syllables can appear in three forms depending on which of the three
> tones are used. These are:
>
> CV? - creaky tone, heavy syllable ending in a glottal stop or
>       glottalized sonorant.
> CVH - level tone, heavy syllable ending in a glottal fricative or
>       a voiceless sonorant.
> CVX - falling tone, heavy syllable with a long vowel or ending in
>       a voiced sonorant.
>
> The result is that all words must have one of the three possible
> stressed syllable forms. Furthermore, stressed heavy syllables vary
> in the coda depending on the tone used. All in all, words without a
> heavy syllable cannot exist. Does this appear natural?
I think it's a not-implausible description of open-class vocab in
English: every word contains a long or closed syllable, and "heavy"
can generalize over "long" and "closed". It should be said, though,
that the syllabifications in mainstream do not support that analysis.

> I'm also not sure if this is a segmental feature alone or a
> prosodic/suprasegmental feature that applies to whole words. Perhaps
> it is both (if that's at all possible). In any case, I'm sort of
> stumped when it comes to representing this orthographically. If word
> stress is almost consistently penultimate and stressed syllables
> consistently heavy, do I still have to represent the coda of
> stressed syllables to mark them as heavy (and thereby also
> stressed)? I was thinking that the type of the coda represented in
> the orthography could be used to mark the tone of the word since
> they are directly related. It just seems gluttonous to overrepresent
> something that is already consistent when tone alone can easily be
> represented by diacritics.
By my reading of your description of Lumanesian, it looks as if there
is a word-level property whereby a word is "+?", "+H", or "+X", which
means that the penultimate syllable is either CV?, CVH or CVX.

Orthographically, then, you need a three-way distinction that is
located in some (ideally unambiguous) position on the orthographic
word (e.g. initially, finally, or on the penult itself). The
distinction nnedn't be marked on the penult if (i) it is marked
elsewhere, and (ii) there is some independent way of delimiting
orthographic words. If one knew the overall distribution of CV?,
CVH and CVX syllables it might be easier to choose between alterntive
orthographic solutions.

A comparison with Livagian may be of interest [especially to you,
because the two conlangs that currently most remind me of Livagian
are Lumanesian and Lojban].
Basically a phonological word in Liv contains exactly one non-low-tone
(either H or HLH) syllable, and any low-tone sylls that precede it in
the word can only have the vowel /oj/ (which in that pretonic position
is subject to additional phonotactic constraints not relevant here). I
have three different orthographic solutions for the 3 different scripts
used to write Livagian.
"syllabary": orthographic words are not delimited, and tone is
    represented as part of the syllable grapheme.
"Livagian alphabet": tone is marked at the start of the orthographic
    word (which is delimited only in this way), and low-tone /oj/ is
    not written.
Roman alphabet: orthographic words are preceded by spaces, non-low
    tone is marked on vowels, low-tone /oj/ is not written.

The Roman alphabet could be used in the same way as the Livagian
alphabet, but the point of using the Roman alphabet is to try to
conform to more international orthographic conventions.

--And.