An acoustic study of the timing of contextual nasalization in English

The time course of contextual nasalization in English was investigated using acoustic data. DFT spectra were analyzed to identify spectral differences betwen contextually nasalized vowels (as in “bean”) and their oral counterparts (as in “bead”). Analysis by synthesis was performed to clarify the contributions of the first formant, the nasal formant, and the accompanying nasal zero, to the spectrum of the vowel. As in previous work, nasalization resulted in spectral changes in the first-formant region, and formants that are shifted relative to what is observed in oral vowels. In our data, the nasal pole falls in the 600- to 1000-Hz frequency range; the nasal zero is usually 200-300 Hz above it. The nasal pole appears 40–50 ms after onset of the vowel (occasionally earlier), and moves up slowly throughout the vowel. These results agree with physiological data in the literature, suggesting that noninvasive, acoustic measures may be useful for investigating the temporal properties of linguistic nasalization.