So I'm assuming the consonant equivalent of a diphthong is a consonant cluster. I'm wondering if these can be interpreted as consonant clusters.

  • ˀm = ʔm
  • dʰ = dh
  • dʱ = dɦ
  • dⁿ = dn
  • ⁿd = nd
  • ᵑd = ŋd
  • ᶮd = ɲd
  • ᶯd = ɳd
  • ᵐb = mb
  • ᵗʃ = tʃ (I saw this along with tʃ as two different elements of a phonology somewhere I think)
  • q͡χ = qχ
  • t͡ɬ = tɬ
  • ʈ͡ʂ = ʈʂ
  • ɖ͡ʐ = ɖʐ
  • t͡ʃ = tʃ
  • d͡ʒ = dʒ
  • ...

I see single symbols such as ʣ, which I assume exist because the consonant cluster [dz] is so common. Please correct me if I'm wrong.

I see these things like [dʰ] sometimes have orthography like "dh", but I'm wondering if the sound would be equivalent to just having a "dh" sound. Same with d͡ʒ = dʒ. The arc there tells me they always appear as a cluster, but wondering if the sound would be equivalent to the two next to each other without the arc.

If not, wondering what the difference is between the superscript and non-superscript versions are.

2 Answers 2


Fundamentally, there's a theoretical difference between coarticulated consonants like /gʲ ⁿd pʰ/ and clusters like /gj nd ph/ rather than a practical one. It's the same as the difference between /t͡ʃ/ and /tʃ/. Namely: in the phonological theory, are these single phonemes, or clusters of phonemes?

In both Japanese and Swahili, for example, we see something that looks like [mb]. In Japanese, this seems to be two separate segments /mb/, since the /m/ for all intents and purposes attaches to the previous syllable, while the /b/ attaches to the following one. But in Swahili, it acts like a single segment /ᵐb/: it appears in places where only a single segment can be, and when you end up with /m/ followed by /b/ in a different morpheme, that's recognized as a different sound (with minimal pairs like mbuni /ᵐbuni/ "ostrich" versus m-buni /mbuni/ "coffee tree").

Like with most things in phonology, it all comes down to which choice makes the analysis simpler and cleaner. There's no one right way to do it, only different analyses.

  • I'd add that you can only answer questions like this on a language-specific basis.
    – curiousdannii
    Oct 10, 2018 at 0:57

I agree with Draconis' answer excluding the details of the claim for Swahili. The fundamental question is whether in Swahili mb 'acts like' a single consonant or like a cluster. The problem is that there is nothing in the grammar of Swahili that clearly diagnoses being a single consonant versus a cluster. There are various clusters of non-Bantu origin like bustani 'garden', γafla 'suddenly', arði 'agriculture', posta 'post (mail)'. Apart from that, there are C+w sequence, mb, nd, ng (also mbw, ndw, ngw) and any sequence of syllabic [m̩] plus C. There is a commonly-invoked "slot" type argument for dealing with supposed consonant clusters, of the form "There are otherwise no clusters of consonants in the onset / coda of the syllable, so if we do X, we can eliminate these putative examples". So, we could say that onsets cannot have any clusters, but then we have to say that the language has added phonemes (ⁿd, ᵐb, mʷ, ᵐbʷ...). So arguments of this type need to be scrutinized carefully, to evaluate the total "cost" of one analysis over the other.

Generally, in Swahili if you have m+C, m is syllabic when C is non-homorganic (m̩tu, m̩levi, m̩geni) or voiceless (m̩pishi). The orthographic sequence mb is ambiguous, hence the pair mbuni 'ostrich' vs m̩buni 'coffee plant', both spelled the same way.

The word mbuni is a class 9 noun, and class 9 has a prefix /N/, meaning that it is a nasal which always agrees with the following consonant. I would argue that it is specifically /ɲ/, since that is the prevocalic form of the class 9 prefix (ɲeupe 'white-9', cf mweupe 'white-1', from /mu-eupe/). The word m̩buni 'coffee plant' derives from mu-buni. We see the stem in the plural mi-buni 'coffee plants'.

The main source of syllabic in Swahili is reduction of /mu/ to [m̩]. The other source is in short words having an initial NC cluster: m̩-bwa 'dog', m̩-bu 'mosquito'. These are class 9 nouns with the prefix /ɲ/, and their nasals become syllabic because otherwise the word would be monosyllabic, and the rules of Swahili avoid that outcome, somewhat. So the nasal becomes syllabic in these words, and then it receives regular penultimate stress (ˈm̩bwa, ˈm̩bu). This in fact creates two other syllabic nasals, [n̩] and [ŋ̩] as in [ˈn̩tshi] 'land', [ˈŋ̩ge] 'scorpion', and while /ɲ/ deletes before voiceless stops (pana 'wide-9' ← /ɲ-pana/, cf. mbaya 'bad-9' ← /ɲ-baya/), it does not delete when it becomes syllabic in [ˈm̩pya] 'new-9' (also 'new-3').

The evidence is pretty strong that non-syllabic clusters in words like ndevu, mbuni, ngano has a bisegmental source: the class prefix /ɲ/ plus a stem consonant. The difference between 'ostrich' and 'coffee plant' lies not in whether there is a prefix, but in what the prefix is and what process creates the CC cluster. My analysis of Swahili is that there are syllabic versus non-syllabic nasals; syllabic nasals are all derived by rule, via (a) reduction of /mu/ or (b) syllabification of a nasal in otherwise monosyllabic words. This is actually consistent with the claim, if one insists on it, that all pre-consonantal nasals which are not syllabic merge with a following consonant, so the clusters nd, mb convert to single segments [ⁿd, ᵐb]. But there is no independent evidence for such a reduction.

I should add that positing [ᵑd] is rather unconventional as a claim about what "a segment" is, since it has differing primary places of articulation. Most people agree to [pl, sp] are consonant clusters, but Osamu Fujimura does not: he claims these are lateralized labials and prespirantized labials. A segment like [ᵑd] challenges the notion of "segment" in the same way. Such challenges are no a priori wrong, but they demand extra scrutiny.

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.