ISCA Workshop on Plasticity in Speech Perception (PSP2005)

Senate House, London, UK
June 15-17, 2005

Acoustic Cue Weighting in the Perception of Initial Voicing in Hebrew Speaking Children and Adults

Riki Taitelbaum-Swead, Minka Hildesheimer, Liat Kishon-Rabin

University of Tel-Aviv, Israel

Studies have demonstrated that while the primary cue for perceiving initial voicing is Voice Onset Time (VOT), listeners may also use secondary cues such as F1 transition and the initial burst. Furthermore studies suggested that the weight assigned to each of these cues for auditory perception may depend on many variables. Therefore it is not possible to directly apply the results stated above to other languages, such as Hebrew, for which different acoustics was demonstrated (even for phonemes common with other languages e.g., ba/pa). In Hebrew, for example, the voiced plosives have a substantial voicing lead and its vowel system includes only five short vowels which may influence the transitions, such as, F1 cutback. This suggests that the cues for voicing perception and the relative weighting may be different in Hebrew from those reported in English. The purpose of the present study was twofold: (a) to examine the relative weighting of some acoustic cues to the perception of initial voicing in Hebrew in different age groups and (b) to compare the influence of vowel context on the perception in children and adults. Three groups of children aged 4-5, 6-7 and 9-10 years and one group of adults, all with normal hearing, participated in this study. Stimuli consisted of naturally produced pairs of meaningful words (/bar/-/par/, /bil/-/pil/). The relative weighting of three cues to the perception of voicing were evaluated: VOT, initial burst, and the transitions. From each pair, the burst and the transitions parts (vowel included ) were extracted, and reconcatenated to produce four new stimuli combinations on a VOT continuum (from -40 to +40 msec). Results show that: (1) phoneme boundaries (PB) of children were least affected by the different stimuli compared to adults, suggesting that children rely more on the VOT cue. (2) the steepness of the slope of the identification curves increased as a function of age, indicating a progressive development in categorization of voicing during childhood, and, (3) children showed different identification functions for the different vowels whereas adults showed similar functions for /a/ and /i/. Thus, the relative weighting of the cues for initial voicing appears to be different in Hebrew than for English. Furthermore, children seem to rely more on temporal than on spectral cues. Finally it is hypothesized that children do not complete the normalization process for speech perception and therefore require separate auditory processing for each consonant- vowel combination.

Bibliographic reference.  Taitelbaum-Swead, Riki / Hildesheimer, Minka / Kishon-Rabin, Liat (2005): "Acoustic cue weighting in the perception of initial voicing in Hebrew speaking children and adults", In PSP2005, 61 (abstract).