HW 03: LINGUISTICS

  1. Record (at 8 kHz) yourself and a member of the opposite sex speaking the words "heed" and "had". Locate the section of the waveform consisting of the vowels. Measure the formant frequencies and bandwidths using a 512-point Fourier transform. Compare and contrast the spectra.

  2. Collect 100,000 words of English text from two different sources. For example, select text from web pages for one sample, and text from a Usenet newsgroup for the second sample. Generate histograms of one, two, and three word sequences. Compare and contrast these histograms. Which words appear frequently in both samples? Which three word sequences occur frequently yet are very domain specific?

  3. Parse these sentences: