Three lO-step [paH~l continua were constructed with F2 as the single place-of-articulation cue. The continua were different in F3, which was rising, level, or falling. Identification tests using these continua were presented to 22 first-, second-, and fourth-grade children. Using logit analysis, individual phoneme boundary values were calculated for each F3 level. It was found that the phoneme-boundary shifts to slightly higher F2 starting values as age increases. Although no definite explanation is available yet, some possible interpretations are discussed.In De Weirdt (1985), age differences in a phoneme boundary were reported. First-grade children (tested at ages 6:2 and 6: 10), and third-and fourth-grade children (mean age 9:5) and adults took an identification test with three-formant stimuli on a The author is "aangesteld navorser" of the Belgian Nationaal Fonds voor Wetenschappelijk Onderzoek. Thanks are extended to M. Vanwonnhoudt, J. P. Martens, and B. Van Coile of tile Laboratory for Electronics and Metrology, for making available the equipment and some software for synthesizing thestimuli, and to G. De Soete and an anonymous reviewer for their helpful advice on an earlier draft of this paper. The author's mailing address is: Dienst voor Psychologie, Rijksuniversiteit Gent, H. Dunantlaan 2, B9000-Gent, Belgium.Maybe younger children are more influenced by the F3 cue-or less influenced by F2-than older ones. One might even speculate that speech categorization in younger children is generally based more on the higher frequency information in the acoustic signal. We were led to this presumption by the results of Irwin, Ball, Kay, Stillman, and Rosser (1985), who demonstrated that auditory temporal acuity develops differently in different frequency regions. They measured the minimum detectable gap in octave-band noises centered around 500, 1000, and 2000 Hz at various ages between 6 and 23 years. Especially between the ages of 6 and 10, the minimum detectable gap decreased, but did so substantially only with the 500-Hz stimulus. Irwin et al. (1985, seeFigure 2) did not cover the whole frequency range, but their data suggest that a general trend may exist. The difference between the level of temporal acuity at age 6 and that of the adult increases as one descends the frequency scale. This pattern might also apply to formants in speech sounds. Assuming that temporal processing aspects are probably crucial for the perceptual effectiveness of formant transition cues, the deviating [pa]-[b] phoneme boundaries of younger children may be due to poorer temporal resolution of the (lower frequency) F2 cue, resulting in higher "weights" for F3.It would be interesting to determine if the F3 cue is more important for younger children. If this were true, however, the developmental shift of the phoneme boundary would be only a stimulus artefact. Accordingly, the earlier study (De Weirdt, 1985) was replicated with stimulus continua at three F3-levels: falling, level, and rising. If younger subjects are more sensitive to t...
