How would one describe the propensity of an aspirate to spread right to an identical consonant in the same word?

  • /tʰeto/ → [tʰetʰo] ~ [tʰeto] (identical consonant); but
  • /tʰepo/ → [tʰepo] *[tʰepʰo] (different consonants)

Can this be construed as a kind of harmony? I never heard of aspiration harmony.

  • 1
    Can you please clarify your transcriptions? As they stand, they don’t seem to make any sense. Why would phonemic /tʰeto/ yield phonetic [tʰeto] and [tʰetʰo] in free variation, but not [tetʰo], while phonemic /tʰepo/ only yields phonetic [tʰepo], not [tʰepʰo] (and no mention of [tepʰo] at all)? May 1, 2023 at 22:08
  • Hopefully this clarifies! If not just say so
    – dOn
    May 2, 2023 at 4:41
  • Okay, so you’re asking specifically about identical consonants? It makes more sense now. I don’t think anyone would expect a reversal of aspiration, though, so those just confuse things. I’ve gone ahead and deleted the reversed options and made the difference a bit clearer. May 2, 2023 at 7:34
  • I'm not a phonologist, but is an "aspiate" a thing, or just a typo? May 2, 2023 at 18:22

3 Answers 3


Some examples from Consonant Harmony: Long-Distance Interaction in Phonology by Gunnar Olafur Hansson (2010):

  • Aymaran varieties, where homorganic consonants have aspiration interactions:

[B]oth Aymaran varieties (Bolivian and Peruvian) appear to disfavor the co-occurrence of plain voiceless stops with either ejectives or aspirates at the same place of articulation. In other words, homorganic sequences of the type /T’…T/, /T…T’/, /Tʰ…T/ and /T…Tʰ/ are extremely rare (MacEachern 1999). Thus, in addition to prohibiting the co-occurrence of otherwise-identical pulmonic vs. glottalic stops — as in Hausa, Yucatec, and Chol — Aymara similarly bans aspirated and unaspirated plosives of the same place of articulation from co-occurring.

  • Gojri/Gujari, a Western Indo-Aryan variety of the Gurjar ethnic group:

Based on a search of Sharma (1979), MacEachern (1999) concludes that in Gojri, homorganic voiceless stops differing in aspiration are not allowed to co-occur within a morpheme (only three exceptions were found). Homorganic voiceless stops that agree in aspiration are allowed (/cʰɪcʰɭəp/ ‘cobra’, /pəɽpəʈo/ ‘blunt’), as are homorganic pairs of an aspirated and unaspirated stop, as long as these differ in voicing (/bapʰəɳ/ ‘eyelash’), but sequences like */tʰ…t/ or */k…kʰ/ are not attested. The Gojri cooccurrence restriction can be analyzed as a root-internal aspiration harmony which is parasitic on identity in both place of articulation and voicing.

  • Interaction of the slack-voiced / depressed plosive series with the aspirated and the plain (although often ejective and sometimes fricated) series in Zulu:

Khumalo (1987) notes the existence in Zulu of a laryngeal consonant harmony governing the morpheme-internal co-occurrence of (non-click) stops: ‘they will either all be [+aspirated] or all will be [+depressed] or all will be unspecified’. The stops Khumalo analyzes as [+depressed] are fully voiced stops (/b/, /d/, etc.), whereas the phonetic realization of the segments he treats as laryngeally unspecified stops (/p/, /t/, etc.) seems to vary between ejectives and voiced fricatives, depending on their position in the word. [...] Khumalo (1987) finds no counterexamples to laryngeal harmony among regular disyllabic roots in Zulu. The phonological reality of the harmony is also supported by data [e.g. /í-kʰôtʰo/ from English court; /úm-bídi/ from English beat] where English word-final /t/ is rendered as aspirated or voiced depending on the laryngeal features of the initial stop.

The case in Zulu and other Nguni Bantu languages has been called laryngeal harmony.


It is claimed to exist in Zulu, but only as a constraint on root-consonant co-occurrence. For example there are roots like [pʰatʰa] 'hold' and [peta] 'dig up', but not [pʰata, patʰa]. Unlike typical examples of harmony, this is only a static constraint no roots and results in no alternations under affixation.

It would be surprising to find such a rule developing, compared to other kinds of harmony, such as nasal harmony, or rounding / front / height harmony. Most harmonies are the phonologicization of local phonetic coarticulation, where e.g. a round vowel to the left doesn't unround fully at the right edge of the vowel so the left edge of the next vowel 100 msc or so to the right maintains an initially lower F2. Aspiration on the other hand is a "done and over" timing relation between a voiceless consonant and a following voiced vowel – the vowel may have an initial short period of voicelessness at the left. In order for long-distance aspiration-spreading to happen at the phonetic level, the vowel between the consonants would have to become completely devoiced, and it is more likely that the vowel would delete entirely (since it would have negligible audibility). Direct aspiration harmony would an unusual non-phonetic sound change, which copies a complex change in laryngeal configuration from one point in the speech stream to another, rather than expanding the time domain of a certain physical state.

  • (I don't know wanything about Zulu specifically but) there could be other ways to explain your Zulu data as well. For example, aspiration may have transphonologized from some vowel quality or other in a parent language. Such a vowel quality could easily smudge from one syllable to the other in the ways you explain in your answer. May 2, 2023 at 14:07
  • As you describe, aspiration is just the devoicing of the left edge of a vowel. But isn’t there also some laxness/tautness of the articulators involved in the production of aspirated/non-aspirated stops? If so, it might be this articulatory aspect of the production which is spreading perhaps. May 2, 2023 at 18:53

I haven't heard of such a thing, but it seems entirely possible.

We know that aspiration can cause dissimilation, as famously seen in Grassmann's Law for Greek and Sanskrit (an aspirated consonant becomes deaspirated before an aspirated consonant in the next syllable; /h/ deaspirates into nothingness).

We also know that it can cause assimilation, if you interpret spellings like Ancient Greek χθ φθ as representing /kʰtʰ pʰtʰ/ rather than /ktʰ ptʰ/ (which many do). In this case, the fact that κ + θ at a morpheme boundary becomes χθ indicates assimilation of aspiration.

And we do see long-distance voicing harmony, where certain types of consonants must agree with each other in voicing. Voicing and aspiration are at their core two sides of the same coin.

So while I can't think of any specific examples of "aspiration harmony", it seems eminently possible, and I wouldn't bat an eye if I saw it in a grammar.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.