Attachments :
 
----- Original Message -----
From: Mark Odegard
To: phoNet@egroups.com
Sent: Tuesday, April 11, 2000 6:27 AM
Subject: [phoNet] Frustration.

Mark writes:
 
Here are some sounds rather difficult to transcribe. Has IPA caught up with the throat-singers of Tuva yet? [...] Scientific American has a nice article on the topic available online:
 

 
IPA has no such ambition. It transcribes speech sounds, not ANY noises that humans can make. There are no IPA letters for a sneeze or a cough, for instance, though special symbols (with appropiriate diacritics) are provided for things like nasalised clicks or velar lateral affricates -- because these sounds DO occur in some languages. As for "throat-singing", it isn't a way of articulating unusual speech segments, but a special technique of voice emission which allows you to sing a tune using amplified vowel formants rather than the fundamental frequency (as in ordinary singing). As each vowel has at least two distinct formants, a throat-singer is able to sing two or even three notes at the same time. I remember that article in SciAm -- the most detailed account of the phenomenon that I've ever read.
 
I think the most competent introduction to the full range of speech sounds in various languages is this:
 
Peter Ladefoged & Ian Maddieson. 1995. The Sounds of the World's Languages. Oxford: Blackwell.
 

[mɑːrk ˈodəgɑrd]

The [d] should probably be a tap.



The above (minus the superfluous length symbol) would do as a phonemic transcription (but then it should be placed in the slanting "phonemic brackets" /.../). Preferably, a stress symbol should be added before "Mark" -- a secondary stress mark if you want to show the prominence pattern of the whole phrase.
 
As for vowel length in American English, I see little point in marking low vowels overtly as long except in stem-final position (as in law, pa), unless you
maintain a difference between bomb (short) and balm (long) (few people do these days). There's a special kind of lengthened "a" as in last, man or bad in most American accents, but it usually differs from [æ] in quality as well, being a diphthong [ɛə] (or even [eə], [ɪə] in extreme cases like New Yorkese).
 
In narrow phonetic transcription you could add details like a tap for /d/ (the first one, of course), the diphthongal quality of /o/, or even the precise quality of the schwa (I may be guessing incorrectly, but I'd expect you to have a relatively high vowel before a following velar).
 
The symbol /r/ used above is what IPA has for a trill; it's routinely used as a cover symbol for rhotics in phonemic transcription (for the sake of convenience); however, since your /r/ is most likely a postalveolar approximant (slightly retroflexed or "bunched-up"), an accurate phonetic transcription should use a more precise symbol:
 
[ˌmɑɻk ˈoʊɾɨgɑɻd]
 
It could be embellished even further.
 
Piotr
 
P.S. An idea has suddenly occurred to me. Unipad text files (.UTX) can be sent as attachments and opened using Unipad. They are not bug-infested like some MS products, and (being just text files) are very small. Could be ideal for sending larger amounts of transcription or any other "special" text. Please check if the one enclosed with this posting opens easily.
 
P.