Consonantal space area in children with a cleft palate: an acoustic study

Marion Béchet, Fabrice Hirsch, Camille Fauth, Rudolph SockInterspeech 2012

TL;DR

This study investigates how children with repaired cleft palates articulate consonants, specifically voiced plosives ([b], [d], [g]), using acoustic analysis. The findings reveal that these children exhibit less precise consonant production and a smaller "consonantal space area" compared to their peers without speech disorders, suggesting compensatory strategies are employed to overcome anatomical challenges.

What Problem Does This Paper Address?

Children born with cleft lip and palate often experience difficulties in speech production due to anatomical differences in their oral cavities, particularly after surgical repair. While previous research has explored vowel production in these children, there's a need for a deeper understanding of their consonant articulation. Specifically, this paper aims to examine the "place of articulation" for voiced plosives ([b], [d], [g]) in children with repaired cleft palates by analyzing acoustic features, and to compare these with control children. The researchers are interested in identifying potential compensatory articulatory strategies that these children might employ to achieve intelligible speech despite their condition.

Methods

The study employed an acoustic analysis of voiced French plosives ([b], [d], [g]) produced by children. Key acoustic parameters, specifically the second and third formants (F2 and F3) at the consonant release, were measured. These formant values are known to correlate with tongue body position and lip rounding, providing insights into the place of articulation. The researchers also calculated a "consonantal space area" for each speaker, derived from the F2/F3 values of the three plosives, using Heron's formula to define a triangular area. Statistical analyses, including ANOVAs, were used to compare acoustic measures between the cleft palate group and a control group, and to examine the effects of pathology and consonantal place of articulation.

Datasets

The study utilized a database of 54 speakers with cleft palate and 26 control speakers without speech disorders. For this specific analysis, 26 speakers with cleft palate were selected. The speakers ranged in age from 9 to 18 years old. The data consisted of recordings of voiced French plosives embedded in carrier words, specifically in V1CV2 sequences where the vowel context was [i] for V1 and [a] for V2 (e.g., "formidable" [ida], "il y balaie" [iba], "la cigale" [iga]).

Key Insights and Ideas

  • Compensatory Strategies: Children with cleft palates appear to adopt compensatory strategies to produce consonants, leading to less precise articulation compared to control speakers.
  • Reduced Consonantal Space: The "consonantal space area," a measure reflecting the distinctiveness of consonant articulations, is significantly smaller in children with cleft palates. This suggests a reduced range of articulatory gestures.
  • Neutralization of Contrasts: There's evidence of a neutralization of the distinction between alveolar [d] and velar [g] articulations in children with cleft palates, indicating they may struggle to achieve the necessary tongue-palate contacts for these sounds.
  • Anterior Articulation Avoidance: Disordered speakers tend to avoid anterior parts of the oral cavity for velar and alveolar productions, possibly due to surgical alterations or velar insufficiency.
  • Variability is Key: Disordered speakers exhibit significantly higher inter- and intra-speaker variability in their acoustic productions, highlighting the persistent challenges in achieving consistent articulatory targets.
  • Hyperarticulation of Vowels: The reduced precision in consonant production might lead these children to hyperarticulate vowels to compensate, a phenomenon observed in previous studies.

Key Results

  • Significant Differences: ANOVAs revealed significant differences in F2 and F3 values between disordered and control speakers, as well as significant interactions between pathology and consonantal place of articulation.
  • F2/F3 Values: Healthy speakers generally had higher F2/F3 values for velar and alveolar sounds and lower values for bilabial sounds compared to disordered speakers. Disordered speakers showed lower F2 values for [d] and [b], and [g] productions had less distinct F2 and F3 values.
  • Consonantal Space Area: The consonantal space area of disordered speakers was found to be significantly smaller (almost half) than that of control speakers (F(1,50) = 12.40, p<0.000927).
  • [d]/[g] Distinction: Disordered speakers showed significantly smaller F3 values for [g] and higher F3 values for [d] compared to controls, suggesting a neutralization of the place of articulation contrast for these sounds.
  • [b] Production: The bilabial consonant [b] was produced comparably by both groups, indicating that lip surgery may effectively address challenges with this sound.

Limitations and Future Work

  • Limitations: The study acknowledges that its findings are based on acoustic inferences and calls for direct articulatory investigations (e.g., electropalatography) to confirm these interpretations. The variability observed, while informative, also makes precise identification of each consonant challenging for disordered speakers.
  • Future Work: The authors suggest extending this study to a larger number of speakers and incorporating direct articulatory data. Further research could also explore the impact of different types of palate surgery on speech production and investigate the acoustic characteristics of CV transitions in more detail. The study also points to the need for further investigation into the specific strategies employed by speakers with different types of cleft palate.

Summary updated

BibTeX
@inproceedings{bechet2012consonantal,
  author = {Marion Béchet and Fabrice Hirsch and Camille Fauth and Rudolph Sock},
  title = {{Consonantal space area in children with a cleft palate: an acoustic study}},
  booktitle = {Interspeech},
  year = {2012},
}
Figure