Speech rates of New Zealand English- and American English-speaking children

Abstract

Adult speakers of New Zealand English demonstrate faster speaking rates than adult speakers of Australian, American, or British English. Little, however, is known about comparability of speaking rates in children who speak different varieties of English. This study begins to address this need through examining articulation rate and overall speaking rate data for ten New Zealand English-speaking children (aged 3;1 – 3;5 years) and ten American English-speaking children (aged 2;11 – 3;5 years). The participants had typically developing speech and language skills and the data were collected from conversational language samples where the child was engaged in a play activity with a speech pathologist or parent. Both groups demonstrated significantly faster articulation rates compared to overall speaking rates. In addition, utterance length was significantly correlated with both speaking rate and articulation rate. Group differences were evident. Contrary to previous results reported for adults, the New Zealand children demonstrated slower overall speaking rates and slower articulation rates than the American children. Results of a vowel analysis indicated a higher percentage of lax vowel production and lower percentage of tense vowel production among the American children compared to New Zealand children, indicating that vowel differences were a contributing factor to the observed differences in speech rates. The clinical applicability of the present data is discussed.

Keywords:

• Articulation rate
• children
• speaking rate
• New Zealand English

Notes

1. Previous investigators of speaking rate in children have either collected a large number of utterances per child across a small subject sample (Hall et al., 1999; Robb et al., 2003b) or collected 15 (or fewer) utterances per child across a large subject sample (Flipsen, 2002; Walker et al., 1992). The present study closely followed the methodology of Robb et al. (2003b).

2. Shriberg et al. (1997) recommended calculation of the percentage of consonant correct-revised (PCC-R) metric to evaluate accuracy of speech sound production. The PCC-R expresses the percentage of intended consonant sounds in a conversational sample that are articulated correctly. The PCC-R considers consonant deletions and substitutions as inaccurate; however consonant distortions are scored as correct.

3. At present there is no consensus regarding the measurement of articulation rate. The lack of consensus centres on the measurement of silent intervals occurring within a speech sample. Some researchers only remove silent intervals of 150 – 250 ms or longer (Flipsen, 2002; Grosjean & Collins, 1979; Hammen & Yorkston, 1996; Tsao & Weismer, 1997). The rationale for including silent intervals of 150 ms or less as part of articulation rate is that these intervals may be associated with articulatory events occurring in the vocal tract during speech production. These silent intervals may be caused by respiratory, phonatory or articulatory processes (Tosi, 1974). However, some researchers believe it is inappropriate to infer articulatory activity on the complete absence of acoustic energy. Silent periods as short as 50 ms may, therefore, be removed from the original utterance (Chen, 1999; Deputy, Nakasone, & Tosi, 1982; Robb et al., 2003a). Measurement of articulation rate according to this latter criterion is considerably more labour intensive in regards to waveform editing. The present researchers chose to remove all silent periods in excess of 50 ms when calculating articulation rate.