The effects of humming and pitch on craniofacial and craniocervical morphology measured using MRI

Traditional voice research occurs within a phonetic context. Accordingly, pitch-related contributions are inseparable from those due to articulator input. In humming, articulator input is negligible. Using magnetic resonance imaging, we test the hypothesis that voice production is accompanied by pitch-related adjustments unrelated to articulatory or postural input.

In this cross-sectional study, 10 healthy volunteers (five men, five women, aged 20-47 years, median 25 years), including singers (6 months to 10 years tuition, median 2 years) and non-singers, were assessed to establish the lowest and highest notes they could comfortably sustain while humming over 20 seconds. With head position stable, midsagittal images were acquired while volunteers hummed these predetermined low and high notes. Twenty-two craniocervical, angular, and linear dimensions defined on these images were compared using one-way repeated-measures analysis of variance. Correlations between variables were sought using Pearson correlation coefficient.

We found significant differences between low- and high-note conditions in six of 22 measures and widespread pitch-related correlations between variables (r≥0.63, P<0.05). Compared with low-note humming, high-note humming was accompanied by increased craniocervical angles opt/nsl and cvt/nsl (P=0.008 and 0.002, respectively); widening of the C3-menton distance (P=0.003), a rise of the larynx and hyoid in relation to the cranial base (P=0.012 and <0.001, respectively), and an increased sternum-hyoid distance (P<0.001).

Voice production is accompanied by pitch-related adjustments that are currently being masked by, or mistakenly attributed to, articulatory or postural input, identification of which could improve understanding of mechanisms underlying speech and song.