Analysis, Vocal-tract modeling, and Automatic Detection of Vowel Nasalization