Re: Status of Italian rising

* Josh Brandt-Young said on 2002-12-09 06:57:57 +0100
> Quoth Mangiat:
> > I can't figure out why linguists tend to describe both the components of
> > falling diphthongs as vocoids (with high /i/ and /u/ lacking sillabicity),
> > while only the second element of rising diphthongs is hold as a vocoid and
> > the first one is described as an approximant (a contoid).
>
> <rant>
> This whole method of describing diphthongs (or even vowels in general)
> irritates me just a bit from a phonetic perspective, because it's at best
> not really descriptive of what's going on, and at worst relentlessly
> inaccurate.
AOL. To model the pronunciation of a language properly, of course the
diphthongs must be taken into account, but it's done badly. I've yet
to see timing-information (length and strength of the segments) of
the diphthongs in a description of a language having them; in fact there's
a general dearth of papers on diphthongs, they're usually explained away
or straight out ignored.

> If I look at a spectrogram of the vowel [i], for instance, I see
> F1 and F2 (first and second formants) at a steady state through the
> duration of the vowel.
That is:

________________ F2

---------------- F1

> For the English diphthong commonly transcribed [AI], though,
> there is no steady state at any point: the formants move smoothly throughout
> the production of the diphthong, starting at [A] and ending at [I], but not
> actually sitting at either of them.
IIRC something like

----____
        ----____ F2

____------------ F1

> In the case of this particular sound, the duration of formant values
> *resembling* [A] is probably longer than those of [I], for which
> reason I suppose it's termed a "falling" diphthong, but really this
> terminology is a bit meaningless.
The fun doesn't end there though; as I've mentioned before (years ago)
the speed at which English AI/aI/Ai (I've seen them all used) falls of
is different to how the speed of for instance the same sound in Norwegian
(which AI/aI/Ai is probably neither rising nor falling). Lets say the
English 'ai' is 2:1 while the Norwegian is 1:1. I had other samples as
well... (looking, not finding). Methinks its time to start up the
spectrogram-makers again and do the good old compare and contrast.


t.