Linguistics L541, Spring 2005
Noah Silbert, Assistant Instructor
Your task is to find a simple phonetic experiment and conduct it. You will work in teams of 2 or 3. Please organize your team by mid February. Each team will design and conduct an experiment together but each student will write his own paper for submission. You may jointly produce figures and tables. Of course, you should discuss the results together and take notes on what you think the data show. Then each student writes his/her own paper. They will be due the last week of class.
The experiment may be many things. I recommend a simple investigation of the acoustic correlates of a phonological or phonetic contrast in some language of the world -- preferably not English. Adam and I will meet with each team before you begin your project to help develop your ideas into an experiment design. When you meet with me, come with several (2-4) experiment ideas, including, if possible, some minimally different test words for each experiment idea. Keep in mind that you will need at least several native speakers of the language and at least one person in the group should be linguistically familiar with the language.
The experiment should be modest in size, requiring no more than, say a 100-200 utterances for analysis. Thus, to provide an example of a typical design, you might have 5 speakers times 8 test words times 3 repetitions = 120 items. From each utterance, you might measure several phonetic variables (eg, formants, segment durations, etc). For more information on designing the experiment and writing it up, check adjacent pages.
Here are a few ideas to suggest the scope of an appropriate experiment.
1. Study the effect of [voice] on vowel duration and syllable-final obstruent duration in English (or German or whatever). Pick, say, 3 minimal pairs (eg, `bigger'-`bicker'). Get 5 recordings of each by 3 speakers (= 6words X 5repetitions X 3spkrs = 90 words). Measure vowel and stop durations (and maybe voice-onset time) for each word. Take means across the speakers; make various displays for each word, probably pooling across the 3 minimal pairs (to look a voicing averaged across the specific consonants and vowels).
2. Effect of `High-Vowel Deletion' (between voiceless obstruents) on mora timing in Japanese. Do the /i/ and /u/ actually disappear (as is often reported)? If so, does this disapperance influence timing of the word as a whole? Find a couple minimal pairs like kakusi and kakesi, (where /u/ should delete but not the /e/); put them in a carrier sentence; make 10-20 recordings each (over several speakers); segment and measure; make visual display of cumulative durations. Then interpret the meaning of your results.
3. Some other ideas:
RFP