Abstract
This study investigates the articulatory dynamics of Hausa ejectives and implosives using real-time MRI (rtMRI) and automated centroid tracking. Results indicate that these non-pulmonic consonants are characterized by highly stable intergestural timing. Ejectives exhibit significant larynx raising with high coupling strength to oral gestures. Conversely, implosives are distinguished from standard voiced stops not by the magnitude of larynx lowering, but by the precise sequencing of the laryngeal gesture relative to oral closure. These findings suggest that the "articulatory choreography" of non-pulmonic sounds is defined as much by temporal coordination as by physical displacement.
Introduction
In linguistics, most speech sounds are "pulmonic," produced using airflow from the lungs. However, many languages, such as Hausa (spoken widely in West Africa), utilize "non-pulmonic" consonants. These include ejectives, where the larynx moves upward to compress air, and implosives, where it moves downward to rarefy air.
While phoneticians have long hypothesized that these sounds require specific larynx movements, direct, quantified evidence from native speakers has been scarce. Previous research often relied on indirect measures, such as external neck sensors or manual tracing of limited video frames. This study addresses the lack of dynamic, automated data to determine whether the distinction between these sounds is primarily a matter of magnitude (displacement) or timing (coordination with oral articulators).
Methods
The researchers utilized real-time MRI (rtMRI) to capture the mid-sagittal vocal tract at a high effective frame rate of 83.3 frames per second. Data were collected from three native Hausa speakers from Northern Nigeria, covering 10 target consonants (ejectives /k’, kʷ’, s’/, implosives /ɓ, ɗ/, and pulmonic counterparts /k, kʷ, s, b, d/). Each speaker produced 168 tokens across phrase-initial and phrase-internal contexts.
To process this data, the study implemented:
- Automatic Centroid Tracking: A novel technique to track the larynx's "center of mass" over time, providing a continuous kinematic profile.
- Region-of-Interest (ROI) Analysis: Used to track oral constrictions (lips, tongue tip, and tongue body) via pixel intensity changes.
- Linear Mixed-Effects Models: Employed to analyze "intergestural lag"—the temporal difference between the onset of oral and laryngeal movements.
Results and Key Insights
Magnitude and Timing Stability
The study revealed that ejectives show significantly larger vertical larynx displacement and higher peak positions compared to pulmonic sounds ($p < 0.01$). Surprisingly, for implosives, there was no significant difference in the degree of larynx lowering compared to standard voiced stops (like /b/ or /d/). Instead, the primary differentiator was timing: in standard stops, the larynx lowers almost immediately, whereas in implosives, the larynx "waits" for oral closure to be fully achieved before dropping.
"Gestural Molecules"
The authors conceptualize speech segments as "molecules" composed of "gestural atoms" (e.g., a tongue movement paired with a larynx movement). The stability of the "bond" (timing) between these atoms is a core feature of the language's grammar. Ejectives exhibited the most stable timing, likely because the massive pressure build-up required behind a sealed mouth necessitates a very narrow window for laryngeal movement. This high "coupling strength" ensures the sound is produced consistently.
Discussion and Limitations
The findings suggest that stability is a defining feature of non-pulmonic sounds. In these sounds, the oral gesture consistently begins before the larynx movement (positive lag), whereas in pulmonic sounds, the movements are nearly synchronous.
However, the study notes several limitations:
- Sample Size: With three speakers, individual variations are present, though broad trends remain significant.
- Place of Articulation: Hausa lacks perfect "minimal pairs" for all non-pulmonic sounds (e.g., there is a bilabial implosive but no bilabial ejective), complicating direct comparisons.
- Aerodynamics: The study tracks movement but does not directly measure air pressure.
Future research may look toward languages like Zulu or Yucatec Maya, which possess more symmetrical sets of non-pulmonic sounds, to further refine models of gestural coordination.
Summary updated