On the Spectrum of Spoken English