Quite recently I had been wondering what the acoustic differences between the two (American) English /r/-allophones ɹ (alveolar approximant) and ɻ (retroflex approximant) are. I downloaded the audio files from the respective Wikipedia articles and analyzed them using Praat. These two shiny formant spectrograms were the result:
In the case of the retroflex approximant the lowering of the third formant was much stronger and more sudden than with the alveolar approximant. Also this low position was maintained for a longer period of time. The second formant moreover seems to have remained more constant.
Those of course are just two audio files, so they are hardly representative.
Does any one of you know if and when you use which sound?