Problems with Recordings

To create my experiment I first recorded my grandmother, a native Thai speaker, pronouncing different Thai minimal pairs. With each word I cut the recording to get the phonemes I was using in my experiment. For example, I cut the recording of the word ‘baa khlang’, meaning ‘insane’, to isolate the initial /b/ phoneme. I then redid this for words beginning with /p/ and /ph/ phonemes. Each phoneme has four separate recordings.

When people naturally speak they change pitch and volume, this natural occurrence affected the uniformity of the recordings. I worried this might cause participants to confuse differences in pitch or volume with differences in phonemes. My concerns proved to be accurate.

At the end of each session I ask them for their initial thoughts on the survey. Some say they struggle to understand some recordings. Others stopped during their survey to ask if a difference in volume means the phonemes are different. T0 limit this confusion before each survey I informed participants that differences in pitch or volume does not mean the recordings are of two different phonemes. I encouraged them to pay attention to the actual sound played.

Since giving my participants a disclaimer I noticed they struggle less with the survey, and finish faster. In the future I will pay better attention to how I cut the recordings to ensure as much similarity in pitch and volume as possible.