Laryngeal characteristics during the production of geminate consonants

Masako Fujimoto, Kikuo Maekawa, Seiya FunatsuInterspeech 2010

TL;DR

This paper investigates the physiological mechanisms behind Japanese geminate consonants, which are known for their longer duration. Using high-speed video and glottograms, the researchers found no evidence of laryngeal constriction. Instead, they observed subtle interruptions in glottal opening and faster vowel cessation during geminate production, suggesting a different articulatory strategy than previously hypothesized.

What Problem Does This Paper Address?

Japanese geminate consonants, often described as "doubled" or "long" consonants, have been a subject of linguistic interest primarily due to their extended duration compared to singleton consonants. While acoustic studies have well-documented this duration difference, the underlying physiological mechanisms, particularly concerning the larynx and glottis, have remained largely unexplored. Some linguistic theories and anecdotal descriptions have suggested a "choked" sound or laryngeal tension during geminate production, but empirical evidence was lacking. This paper aims to bridge this gap by providing a physiological examination of how geminate consonants are produced, specifically looking for evidence of laryngeal or glottal constriction.

Methods

The researchers employed a high-speed digital video recording system capable of capturing images at 4500 frames per second. This allowed for detailed visualization of laryngeal and glottal movements during speech. Simultaneously, audio signals and photoelectric glottograms (PGG) were recorded. PGGs measure the light passing through the glottis, providing a quantitative measure of glottal area changes over time. The study analyzed these recordings to compare the glottal width and movement patterns during the production of singleton consonants versus geminate consonants.

Datasets

The study used custom-generated nonsense words in Japanese, consisting of six sets of words, each containing a singleton and a geminate consonant. For example, the /te/-set included "etete" (singleton /t/) and "ette" (geminate /tt/). These words were spoken by a single male native Japanese speaker in his mid-30s. The paper does not introduce a new benchmark dataset but rather utilizes a controlled set of stimuli for analysis.

Key Insights and Ideas

  • No Laryngeal Constriction: Contrary to some hypotheses, the high-speed video analysis revealed no apparent constriction or tenseness in the larynx or glottis during the production of geminate consonants. This suggests that the "choked" or "tense" descriptions might be auditory impressions rather than direct physiological actions.
  • Subtle Glottal Width Differences: While not dramatic, geminate consonants showed slightly wider glottal openings compared to their singleton counterparts. This width variation was found to be consonant-type dependent, with fricatives exhibiting wider openings than plosives, and palatal/palatalized consonants showing wider openings than non-palatal ones. This is likely related to the need for stronger noise generation in these consonant types.
  • Articulatory Timing Matters: The PGG analysis revealed two key differences: an interruption or dip in the glottal opening movement during geminate production, and a faster cessation of voicing for the preceding vowel. These findings point towards timing and movement efficiency as crucial factors in geminate consonant articulation, rather than a simple increase in laryngeal tension.
  • Fricatives are Different: The observed glottal opening interruptions and faster vowel cessation were not consistently found in fricative geminates (/ss/, /ssh/). This is attributed to the continuous nature of frication, where complete oral closure is not required, thus potentially altering the laryngeal coordination.

Key Results

  • High-speed video analysis showed no visible laryngeal or glottal constriction during geminate consonant production.
  • Glottal width for geminate consonants was slightly wider than for singleton consonants, with variations based on consonant type (e.g., wider for fricatives than plosives).
  • Photoelectric glottograms indicated that geminate consonant production can involve an interruption in the smooth glottal opening movement and a quicker cessation of the preceding vowel's voicing, particularly for plosives and affricates.
  • These specific PGG findings were less apparent in fricative geminates.

Limitations and Future Work

The authors acknowledge that this study involved only one subject. They emphasize that additional data from more subjects and further examination are necessary to confirm these findings. Future work should focus on detailed analysis of the initial phases of geminate consonant production to further elucidate their characteristics. Investigating a wider range of consonants and languages would also be beneficial.

Summary updated

BibTeX
@inproceedings{fujimoto2010laryngeal,
  author = {Masako Fujimoto and Kikuo Maekawa and Seiya Funatsu},
  title = {{Laryngeal characteristics during the production of geminate consonants}},
  booktitle = {Interspeech},
  year = {2010},
}
Figure