Note: you can skip Section 5 for now; we'll get into that later.

0. If there are any general sections/topics/concepts you don't understand or think you are missing background to understand, please point out what these are.

1. Suppose I want to study the voice quality of the word /iʔa/ in a language (this is the word for `fish' in Samoan). If this word is phonetically realized with a glottal stop [iʔa], can I measure spectral parameters such as H1-H2 during the glottal stop? Why or why not? If not, what recourse do I have?

2. Consider /b/ vs. /p/. Which would you expect to have greater vocal fold approximation and why?

3. Consider the use of the term "dimension" in the discussion and presentation of Table 1. How this use of "dimension" as a term compare to the use of "dimension" as a term in Table 5, or in Figure 3? Do you think it'd be appropriate to swap out the term "parameters" in Table 3 for "dimensions"? Why or why not?

4. According to the paper, how is vocal fry different from prototypical creaky voice?

5. Page 11 notes that "the source spectrum model is therefore used only for f0 values below 500 Hz; higher than that, and H4 will be equal to or surpass 2 kHz!). Explain why it is the case that if f0 > 500 Hz, H4 will be equal to or surpass 2 kHz.

6. Describe in plain English what the following parameters measure: (a) H1-H2, (b) H1-A4.

7. Consider the spectrum shown in Figure 4. Is H1-H2 positive or negative? What about H2 kHz- 5kHz? Why?

8. Consider Figure 6. Note that the sounds from which the spectra were taken were synthesized. Let's assume that everything in the synthesis was constant other than vowel quality parameters. Estimate what H1-H2 is in each of the sounds. Does it differ between them? Would you expect it to? Why or why not?

9. Suppose we have Vowel X and Vowel Y. We find that HNR is higher in Vowel Y than Vowel X. Which vowel is noisier and why?