Setting Up


In the next part of my research, I took the different consonant and vowel parts and crossed spliced them within their group. For example, a lenis [t] consonant part might be paired with an aspirated [t] vowel part, or a fortis [s] consonant part would be paired with a lenis [s] vowel part. This produced 22 stimuli, including the original unspliced words. These unspliced words were kept in as a sort of control. If listeners couldn’t correctly identify these normal Korean words, then it is doubtful that they will provide any useful information about the hierarchy of cues of these consonants. Since I was performing this experiment in the United States, many of my participants were Korean Americans. Their ability to talk and understand Korean varied greatly depending on whether they used mostly Korean or English when they were young. This means that it was important to keep this kind of control in the experiment.

After I made and manipulated the stimuli, I met with my advisor again to determine how I could set up my experiment. Originally, I was hoping to create a webpage that my experiment could run on, so that I would be able to send the link to a large number of people in both the US and Korea. This would have made it much easier to gather a large amount of data. However, this proved to be much harder than anticipated. I was not able to find a platform that allowed me to set up the experiment in the way that I desired. All of online platforms that I found did not have a feature that kept sound clips from being played more than once. The websites also did not allow a clearness scale, which I had hoped to add.

However, my advisor, Professor Lunden did show me how to set up a Forced Multiple Choice test using Praat. It was her idea to add a clearness scale, so that participants had to indicate the clearness of each stimuli they heard. It if was difficult to understand the word, they choose a 1. If it was very clear, they choose a 5. If it was somewhere inbetween, they could choose a number between 1 and 5 on the scale. This will help us better understand which cues in the words are strongest. It may be possible that some cues override all the others quite strongly, so the participants will be able to understand the word easily. However, some conflicting cues may instead cause the words to be very difficult to indentify one way or the other. I had each stimuli play 3 times in a random order, so that there were 66 responses for each participant.