--- In
cybalist@yahoogroups.com, "Richard Wordingham"
<richard.wordingham@...> wrote:
> Does the contrast of [j] and [i]
> ever have much functional yield?
> The constrasts in English, e.g.
> <ear> via slurred <you>, are much
> more a matter of stress placement.
> The yield is pretty low (close to
> zero) in Latin, but who argues that
> it had 3 vowels rather than 5?
> Richard.
Maybe /i/ _should_ be analyzed as an allophone
of /j/ in English. If <ear> is analyzed as
beginning with a glottal stop, and <you> as
with a /j/, then you have no problem. Jens
describes a similar necessity for inserting
just such a "shadow consonant" in his post at
http://groups.yahoo.com/group/pieml/message/406
where he writes "... if we introduce a "shadow
consonant" /H/ which acts like a laryngeal in
that it lengthens a preceding syllabic and
influences the syllabification of sonants.
In that case /vrHa-/ may in fact be found
to yield ura-, while /vra-/ stays unchanged,
by absolutely consistent rules. A "shadow
consonant" is of course needed anyway in order
to account for certain word forms to behave
normally in sandhi. We then also get a lot
of inflectional oddities to come out exactly
as expected."
David