Word or Phoneme? To Optimise Prosodic Features to Predict Lung Function with Helicopter Task

The study aimed to decide whether word-based or phoneme-based acoustic features could significantly correlate to prosodic and lung function measures. "Speech breathing" usually means producing the airflow required for phonation by utilising expired air and lung mechanics. Voice analysis as a health indicator has been extensively documented. The "helicopter task" is an articulatory test that measures the endurance of the respiratory system by having participants quickly repeat the word "helicopter" for three 20-second runs, separated by two 20-second silent, relaxed breathing intervals. Ten native English speakers' speech data that correlated with lung function measurements were used in the study. The study used ten native English speakers' speech data correlating to lung function measures. Specifically, both word-based (“helicopter”) and phoneme-based (e.g., fricative consonant /h/, plosive consonant /k/, /p/, /t/) prosodic analysis was run to correlate to speech rate, word duration and lung function measures. Furthermore, the run effect on prosodic features at word and phoneme levels was investigated. The study found that, among nine phonemes in the word helicopter, /h/, /p/, /ɔ/ and /ə/ were significantly correlated with speech rate and word duration. In addition, it was found that the plosive phoneme /p/ duration became more variable in the third articulation run than in the first and second runs. It was explained that consonant /p/articulation change might reflect the taxed and exhausted respiratory system when the task was carried out. The consonant /p/ might be the best phoneme candidate to replace the whole helicopter to predict lung function measures.