Ejectives and Implosives

How the glottalic airstream works

Most consonants in most languages are pulmonic — air flows outward from the lungs through the larynx and out the mouth or nose. Ejectives and implosives interrupt this flow. They close the glottis (so no lung air can pass) and use the entire larynx as a moving wall to compress or rarefy a small pocket of air trapped above it.

The articulation has three coordinated events:

1. Oral closure. The lips, tongue tip, or tongue back close off the vocal tract at the same place as a corresponding pulmonic stop ([p], [t], [k]).
2. Glottal closure. The vocal folds shut tight, sealing the larynx. Now a sealed pocket of air sits between the closed glottis and the closed oral place.
3. Glottal piston motion. The entire larynx moves vertically. Up — air compresses (ejective). Down — air rarefies (implosive). When the oral closure releases, the pressure differential drives the airstream.

For ejectives, the larynx rises 1-3 cm during closure, raising intraoral pressure from atmospheric to 4-9 kPa. When the lips open, air explodes outward in a sharp burst. There is then a silent gap — the glottis must descend back to phonation height before voicing of the next vowel can begin. For implosives, the larynx lowers, and ambient air pressure pushes inward when the closure releases. The vocal folds vibrate weakly throughout (powered by lung air leaking past loosely held folds), giving implosives their characteristic voicing.

Why glottalic consonants matter

• Phonological typology. The glottalic airstream is one of three universal mechanisms (pulmonic, glottalic, velaric); every language uses at least one.
• Acoustic phonetics. Ejective bursts and implosive pressure trajectories are acoustically unique — diagnostic in fieldwork.
• Historical reconstruction. Glottalized stops in Proto-Indo-European have been proposed (Gamkrelidze and Ivanov 1973) — a controversial revision of the traditional reconstruction.
• L2 acquisition. Most adult learners of glottalic-rich languages need explicit instruction; mastery is gradient, not categorical.
• Speech pathology. Glottalized stops mark several disorders (apraxia, dysarthria) in speakers of non-glottalic languages.
• Sign-and-speech parity. Glottalic articulations are part of the universal phonetic inventory hypothesized to constrain spoken-language phonology.

Glottalic vs pulmonic stops

Feature	Pulmonic [p]	Ejective [pʼ]	Implosive [ɓ]
Airstream	Pulmonic egressive	Glottalic egressive	Glottalic ingressive
Glottis state	Open or vibrating	Closed, raised	Closed, lowered (vibrating loosely)
Intraoral pressure (peak)	1-2 kPa	4-9 kPa	Below atmospheric
Burst duration	15-30 ms	Under 10 ms	30-50 ms
Voicing during closure	None (voiceless) or full (voiced)	None	Weak, declining
VOT (next vowel)	0-100 ms (varies)	50-150 ms (long silent gap)	Voicing continuous from closure into vowel
Spectral peak at release	1-3 kHz	4-8 kHz (sharp, high)	0.3-1 kHz (low)
Cross-linguistic frequency	~100% of languages	~18%	~13%

The acoustic differences are striking enough that ejectives and implosives can be reliably identified from a spectrogram alone, without articulatory data — useful in fieldwork where the analyst cannot palpate the speaker's larynx.

Cross-linguistic examples

Quechua (Andean South America) contrasts plain, aspirated, and ejective stops at five places: /p pʰ pʼ/, /t tʰ tʼ/, /k kʰ kʼ/, /q qʰ qʼ/, plus affricates /tʃ tʃʰ tʃʼ/. The minimal triple kara ("skin"), kʰara ("expensive"), kʼara ("salty") differs only in laryngeal feature. Cusco Quechua speakers reliably produce all three categories from age four.

Hausa (West African Chadic) contrasts plain voiced /b d/ with implosive /ɓ ɗ/ — but only at bilabial and alveolar places. There is no implosive velar in Hausa (a typological gap also seen in Vietnamese and Khmer). The minimal pair baba ("father") versus ɓaɓa ("large") is foundational in Hausa phonology coursework.

Sindhi (Indus Valley) has the largest implosive inventory in the world: /ɓ ɗ ʄ ɠ/ at four places of articulation, contrasting with plain voiced /b d ɟ ɡ/ and aspirated /bʱ dʱ ɟʱ ɡʱ/. Sindhi children acquire implosives early; the four-way contrast is robust in connected speech.

Georgian (Caucasus) has /pʼ tʼ kʼ qʼ tsʼ tʃʼ/ contrasting with /p t k q ts tʃ/ and aspirated counterparts. The Georgian word kʼali ("woman") versus kali ("calorie", borrowed) is a familiar pedagogical pair. Georgian also exploits ejectives in famous tongue-twisters that string four glottalic stops in a row.

Zulu and Xhosa (Bantu) have both ejectives ([pʼ tʼ kʼ tsʼ]) and a single implosive ([ɓ]). The ejective series contrasts with aspirated and breathy-voiced stops, giving Zulu a four-way laryngeal system.

Worked examples — the airstream in action

Producing an ejective [kʼ]. Try this in a quiet room. Close the back of your tongue against the soft palate, as for [k]. Without releasing, swallow — feel your larynx rise. Now, with the larynx still elevated and the glottis sealed, release the tongue closure suddenly. You should hear a sharp pop, much sharper than ordinary [k]. The trapped air, compressed by the elevated larynx, escapes in a high-pressure burst. Quechua and Georgian speakers do this in roughly 100 ms, between vowels of normal length.

Producing an implosive [ɓ]. Close your lips for [b], but don't push lung air outward. Instead, lower your larynx — feel it descend. The vocal folds should vibrate weakly during this descent (powered by ambient lung air leaking past loosely held folds). Now release the lips. Air should rush inward for a brief moment before normal voicing resumes. The result is a soft, low-pitched [ɓ]. Hausa speakers produce this so quickly that the inward airflow is rarely audible — only the spectral signature remains.

Pulmonic vs glottalic burst. Record yourself saying English "pa" and a Quechua speaker saying "pʼa." Open both files in Praat. Measure the intensity of the burst (the brief energy spike at release). The pulmonic [p] burst typically peaks around 70 dB SPL; the ejective [pʼ] peaks around 80-85 dB. Now measure the spectral centroid of the burst — pulmonic [p] centers around 1-2 kHz, ejective [pʼ] around 4-6 kHz. The "crisp" perceptual quality is exactly this high-frequency intensity peak.

Variants and edge cases

• Ejective fricatives. Tlingit, Amharic, and several Salishan languages have [sʼ ɬʼ ʃʼ] — fricatives produced with a glottalic egressive airstream. The articulation is a brief frication followed by glottal release. Acoustically the fricative noise is shorter and more intense than pulmonic counterparts.
• Voiceless implosives. Rare but attested in Owerri Igbo and some Bantu languages. Transcribed [ɓ̥ ɗ̥]. The larynx descends but the vocal folds do not vibrate. The acoustic signature is similar to a creaky-voiced stop with low-frequency burst.
• Ejective approximants. Even rarer — described for Yapese and a few others. Contested whether these are true ejectives or simply glottalized approximants.
• Glottalized sonorants. Many North American languages have [mˀ nˀ lˀ wˀ jˀ] — sonorants with a partial glottal closure. These are not strictly ejectives (the glottalic airstream mechanism doesn't apply to sonorants), but they share the glottalization feature.
• Pre-glottalized stops. Some languages (Vietnamese, Burmese) have [ʔp ʔt ʔk] — glottal closure precedes oral closure. These are sometimes confused with implosives in transcription but are articulatorily distinct.

Common pitfalls

• Confusing ejectives with aspirated stops. Both have a long silence after release. Ejectives have a high-frequency, brief burst; aspirated stops have noisy, longer aspiration.
• Confusing implosives with prevoiced stops. Spanish [b] has voicing during closure (-100 ms VOT) but pulmonic egressive airstream. Hausa [ɓ] has voicing and ingressive airstream. Look at intraoral pressure trajectory to distinguish.
• Treating the IPA apostrophe as ASCII. The ejective marker is U+02BC, not the typewriter apostrophe. Software that auto-corrects can silently corrupt transcriptions.
• Assuming all languages with glottal stops have ejectives. Many languages (Hawaiian, Indonesian) have [ʔ] without any ejective contrast — glottal stop is pulmonic-suspended, not glottalic.
• Treating ejective and "glottalized" as synonymous. Glottalization is a broader feature; ejective is a specific airstream. Glottalized sonorants are not ejectives.
• Assuming high-altitude correlation is universal. Caleb Everett's altitude-ejective hypothesis (PLOS ONE 2013) has been challenged — the correlation may be confounded by language family clustering in highland regions.