Re: Stress and vowel length in Tirelat

Benct Philip Jonsson wrote:
> In Swedish vowel length, stress and syllable coda weight are bound up
> with each other in complicated ways. The only things everyone is
> agreeing on seem to be that stress placement is distinctive and that
> fully unstressed vowels can't be long. In particular it is
> controversial whether vowel length is distinctive or a function of
> stress and syllable structure. I'm nowadays tending to believe that
> dialects and idiolects differ in this regard. For myself vowel length
> is clearly automatic, and I have trouble coping with distinctive vowel
> length in other languages. Thus my immediate thought was that Tirelat
> vowel length is a function of distinctive stress, but that certain
> coda types attract stress and/or cause length, and thus I'd write
> [ma'ra:t] as _marát_, and in general mark irregular stress and/or
> length with an acute. However the thought hit me that apparent long
> vowels might be sequences of vowel + semivowel and/or vowel + /h/,
> supposing a sequence [h] > [h\] > [:], or vowel + /h\/, /?\/ or /?/ if
> there are (or were) such phonemes. Note that voiced laryngeal or
> pharyngeal fricatives would be stress attractors according to the
> rules you gave. Does the language have a /?/ phoneme already? It has
> been suggested that French h aspiré and e muet are in fact a single
> phonemes with the contextual and stylistic allophones [?], [@] and
> zero!
Tirelat has two semivowels /w/ and /j/; earlier versions also had /H/,
and from time to time I've considered bringing it back along with the
voiceless approximants. Note that what I'm calling diphthongs /aĭ/, /oĭ/
etc. could be described as /aj/, /oj/, and the rule might simply be that
stress is attracted to syllables ending in a voiced consonant. If there
used to be another voiced approximant, e.g. /M\/, it may explain the
vowel length in unusual words like "pasiraa" (cucumber). If /ma'ra:t/
has a similar history, though, there ought to be words like *maraĭt or
*maraŭt.