Acoustic characteristics of Cantonese tones produced by adults and children

Abstract

Large discrepancies have been reported on the age of acquisition of lexical tones. To date, no study on children’s Cantonese tone production has provided detailed acoustic analysis to support the perceptual findings. This study aimed at investigating the acoustic characteristics of children’s Cantonese lexical tone productions. Nineteen three-year-old children and their mothers were recruited. They were asked to produce monosyllabic Cantonese words in isolation in a picture-naming task. The productions were low-pass filtered and categorized by five judges. Fundamental frequencies and slopes of the tones in adults’ correct productions, children’s correct productions, and children’s major tone errors were compared. The findings showed that most of the children’s correct productions were acoustically similar to those of adults. However, in T1(HL) and T2(HR), even children’s tones that were correctly categorized by the judges were acoustically different from those of adults. Children’s major tone errors were phonetically different than adults, and the acoustic parameters of most errors matched the acoustic characteristics of the perceived tones, supporting the judges’ perceptual categorization of the tones. The findings provided additional evidence that three-year-old children have not produced adult-like tones and the use of low-pass filtering to control for lexical biases in tone judgment is valid and reliable.