Dr. Deka thanks for the reply.
There are wide linguistic differences within Assam and
in such circumstances it is always dangerous to make
sweeping generalizatoins. Therefore, it is good that
you're taking up the matter.
I'm from Jorhat and when I speak I never get an uvular
effect, more so with front vowels in words like /xex/.
In the rcordings of Dr. Deka, the /x/ there definitely
sounds uvular. I had noticed with a some sounds of
Kamrupi. For instance, I always thought the /k/ of
Kamrup/Nalbari ('velar voiceless plosive'to go by
phoneticky descriptions)is much 'backer' than the one
produced by people from my part of the state.
Due to the pressure of other work I havn't been able
to spectrally analyze Assamese sounds recently. But I
think it will be a good project to compare the
different /x/ s of Assam. A colleague of mine here in
Holland is involved with a huge project which records
the varieties of /r/ in many varieties of Dutch in
Holland and Belgium (apparently there are 36 which
also includes a /x/!)
Regards,
Shakuntala
--- rabindeka <[EMAIL PROTECTED]> wrote:
>
>
> appreciate Shakuntala for the response!
>
> [I, for instance, as a native speaker, find it
> difficult to swallow
> that the /s/ or /x/ (whatever you want to call it)
> are two different
> phonemes in Assamese in the word /xex/, as they are
> not categorially
> different.] re:
>
http://groups.yahoo.com/group/assamonline/message/2363
>
> sorry Shakuntala! :)
> one would probably appreciate that i did not tell
> about this findings
> for over a decade, but seeing IIT Guwahati talking
> about it i felt i
> needed to tell.
>
> the following are two experiments to explore (will
> publish the
> results of both experiments in a subsequent email).
>
> thanks,
>
> rabin
>
>
>
> DATA PREPARATION:
>
> /xex/
>
> lets give a name for first /x/ form left as S01 and
> second /x/ as
> S03. so we call /xe/ equals S01E, and /xex/ equals
> S01ES03.
>
> Now lets take the word /xaxadhar/, i.e., sasadhar.
> in sasadhar, /sa/
> equals S01, and lets define /dha/ equals DH01 and
> /r/ equals R01.
>
> so we can write S01S01DH01R01 instead of sasadhar or
> xaxadhar.
>
> lets record DH01, R01, S01, S01E and S03
> respectively ( i have
> recorded DH01.wav, R01.wav, S01.wav, S01E.wav and
> S03.wav that can be
> downloaded form link
> http://www.geocities.com/rabindeka/data/phonemes/ )
>
> EXPERIMENT 1:
>
> xex = S01ES03 -------------------------> (1)
> sasadhar = S01S01DH01R01 --------------> (2)
>
> lets set S01 = S03 so the equation (1) and (2) can
> be rewritten as
> follows
>
> xex = S03ES01 -------------------------> (3)
> sasadhar = S03S03DH01R01 --------------> (4)
>
> now what will you hear for playing the sequence,
> i.e., S03ES01, in
> Equation (3)? will you hear same, i.e, S01ES03 as in
> Equation (1)?
>
> similarly,
> what will you hear for playing the sequence, i.e.,
> S03S03DH01R01, in
> Equation (4)? will you hear same, i.e, S01S01DH01R01
> as in Equation
> (2)?
>
> Note:
> in order to create a word from phonemes, one would
> need sophisticated
> digital signal processing (DSP) tools. such tools
> are not cheap,
> e.g., MATLAB is a seven thousand dollar tool set. so
> you would need
> to rely on observation.
>
> EXPERIMENT 2:
> the PRAAT tool is developed using digital signal
> processing
> principles/algorithms (though no way to validate its
> performance
> accuracy). this tool can be used to assess the
> frequency bandwidth
> for S01.wav and S03.wav (you might as well record
> your won voice for
> S01 and S03 respectively).
>
> use PRAAT to measure frequency-bandwidth
> for S01.wav and S03.wav -------------> (5)
>
> now use fundamental science in order to find
> frequency-bandwidth for
> human voice, then compare this frequency-bandwidth
> with the results
> found in Equation (5).
>
> SUMMARY:
> the above two experiments would enable one to assess
> whether S01
> equal S03 or not. the second experiment will give a
> scientific
> account whether S01 and S03 are voiceless or not.
>
> WHAT WILL HAPPEN IF ONE EQUATE S01 to S03?
> a range of voice-activity related equipment/product
> will fain that
> are used in digital signal processing discipline in
> general --
> including telecommunications, wireless, voice over
> IP, speech
> recognition, biomedical ... --- this is what is
> something IIT
> GUWAHATI should talk!
>
>
>
>
>
>
>
>
Shakuntala Mahanta
let.uu.nl
Trans 10,
3512 Jk,
Netherlands
Tel 031-30-2536052(off)
Fax 031-30-2536000(off)
