Through these exercises, you will practice general and impressionistic transcription (Section 1), measuring vowel formants (Section 2), estimating the length of your vocal tract (Section 3), measuring properties of stops (Section 4), and measuring properties of fricatives (Section 5). Don't forget the reference provided previously, the basic guide to using Praat that I put together. You may use Praat for any of the sections, and you will likely need it for most of them (unless you are using other, equivalent software).
Provide IPA transcriptions of the following words as they're pronounced in the recording, accessible from Moodle. Use a narrow transcription for words of American English (1-10), and an impressionistic transcription for words in other languages (11-15). Three repetitions are provided of each word.
Be as accurate as possible, given that you can't know exactly what the articulators were doing during the recording. Be careful not to base your English transcriptions on the orthography or your own pronunciation. By nature, three different utterances of the "same" thing will always be a little different; in this case, while not intended during the recording, some of these differences may be large enough that you'll want to transcribe the utterances differently. If so, make note of what was different, and provide your best transcription of the different utterances. If your pronunciation of the English words is different, feel free to provide a transcription of your pronunciation—though this is optional.
Record yourself saying five different one-syllable English words containing the vowel phonemes /i ɪ ɛ ʌ ɑ ʊ u/ (being careful not to use words with /ɒ/ for /ɑ/ even if you merge them). Choose one-syllable words with only voiceless sounds around the vowel to make it easier to see the vowel. For example, you could use the words seat, feast, thief, seep, and quiche for /i/.
For each word, measure F1 and F2 over a stretch of at least 20ms in the center of the vowel. Keep track of these measurements in a spreadsheet program. Determine the average F1 and F2 values for each vowel phoneme using the measurements for each group of words. Also determine the standard deviation from the mean of the measurements. For each value, you should have a table that looks roughly like the following:
|/i/||F1 (in Hz)||F2 (in Hz)|
|mean ± stddev|
|269 ± 17||2263 ± 81|
When you are done collecting all the measurements, plot the means of all of the vowels in an X/Y scatter plot. That is, each vowel should be plotted as a point (labelled with the IPA symbol for the vowel phoneme) in the plot. Make sure you have the axes "reversed" from the the default, so that F2 increases from right to left, and F1 increases from top to bottom, recreating the orientation of the IPA chart.
Submit your spreadsheet and your graph separately; the graph should be in a graphics format of your choice (PDF or PNG preferred). Also submit your recordings.
Review the page about calculating the length of your vocal tract (posted on Moodle) from Kieth Johnson's Phonetics textbook.
Try to find your neutral vocal tract position as the text suggests, including recording words that start with /ə/, and by recording a range of vowels.¹ When the difference between F1 and F2 (F2-F1) and the difference between F2 and F3 (F3-F2) are about the same, you've found a neutral vocal tract configuration.
Use these values to calculate the length of your vocal tract. Specifically, perform the calculation three times: once using each of the first three formants (F1, F2, and F3). Provide the formant values and the lengths you arrived at. Assume that the speed of sound (c in the equation) is approximately 35,000 cm/second.
Are the resulting measurements about the same? If they're vastly different, which seems more accurate, and why (and why are they so different?)? If the resulting figures are about the same, do you think they're are about right, given the average length for men reported in the course packet, the fact that vocal tract length correlates to height, and the fact that (if relevant) women typically have shorter vocal tracts than men?
¹ Note: if you think you might be sensitive about the length of your vocal tract and/or height, feel free to find a friend (who my remain anonymous, but must not be a classmate) to work with to make the recordings.
Record yourself saying the following words, one repetition each. If you're not a native speaker of American English, you may either find a native speaker or just record yourself—it's up to you, but let me know what you did.
Measure the voice onset time (VOT) of the orthographic ‹t› or ‹d› in each word (just the ones that are bolded), from the beginning of the stop's release to the beginning of voicing (the first glottal plus, or blue dot if you have Pulses → Show pulses on). Negative VOTs will show up as a voicing bar in the period of silence, positive VOT values will show up as periods of aspiration after the release burst, and near-zero VOT values will have no voice bar, and voicing will start basically immediately after (or during) the release burst.
In all but the initial position (i.e., in 2, 3, 4, and 6), measure the closure duration of the alveolar stops.
For the above measurements, record your measurements in a spreadsheet, with the words in the first column, VOT in the second column, and closure duration in the third column. In a separate sheet (in the same file), for each group of three words, find the mean VOT and standard deviation, and mean closure duration and standard deviation. This second sheet should have 5 columns: group of words (by number), mean VOT, standard deviation of VOTs, mean closure duration, standard deviation of closure durations. Highlight the highest means in bold and the lowest means in italics. Turn in your spreadsheet file and recordings.
Record yourself saying the following words, 3 repetitions each:
Make a spreadsheet with each repetition (fin #1, fin #2, etc.) in the first column, the duration of each fricative in the second column, and the center of gravity of the fricative in the last column.
To measure center of gravity in Praat, first go to File → Extract selected sound (time from 0) while the entire duration of the fricative is selected. Then go to your objects list, select the new sound object, click Analyse spectrum → To spectrum. Select the new spectrum object, and click Query → Get centre of gravity..., and use a power of 1.
As with stops, make a separate sheet in your spreadsheet, listing groups in the first column (in this case, /f/, /θ/, etc.), average duration in the second column, standard deviation of duration in the third column, average center of gravity in the fourth column, and standard deviation of center of gravity in the fifth column.
Turn in your recordings and spreadsheet, and discuss your results, including any notable patterns and how the measured properties of the sounds compare to our expectations.