I have this simple question on Kiswahili, a Bantu language.

As you know in english, we can not always define morae. it's completely different from Japanese morae system.

But when I learn Swahili, sometimes there appears the term morae for the explanation.

So, is swahili (and many other Bantu languages) actually mora-counting language??

I mean, I wonder if we can always define morae consistently in Swahili or not... It seems like the concept of morae is required to describe accent patterns in swahili, in my opinion. But how do you guys think?

  • 1
    ...or maybe I'm just interested in whether morae are 'significant' in Swahili or not?
    – mt.tread
    Commented Feb 28, 2019 at 4:32

2 Answers 2


In essence, Swahili stress has two rules:

  • If the word is shaped like NC(C*)V, the first nasal is syllabic, and stressed. (For example, ḿbwa "dog", ḿtu "person".)
  • Otherwise, the stress is on the second vowel from the end, and there are no syllabic nasals. (For example, kusóma "to read", nitazibembeléza "I'm going to pet them", kondóo "sheep".)

One way to analyze this, like you said, is morae. Nasals before other consonants count as one mora, and every vowel written down gets one mora. The stress goes on the second mora from the end, and if a nasal takes the stress, it turns syllabic.

However, you could also handle this without mora-counting at all. "Nasals before other consonants" are always either stressed and syllabic, or can be analyzed as prenasalization rather than as consonants of their own. So you might analyze mtu as being underlying /mətu/, perhaps, with a rule that makes schwa disappear (and NC sequences become prenasalized) whenever it's not stressed. This would also explain all the data nicely.

So, short version: as with everything in linguistics, there's no hard-and-fast right answer. Use whichever model you think explains the data most accurately and elegantly!


The term "mora counting" has a specific technical meaning, based on the connection between duration and certain linguistic units (moras, syllables, stress feet). In a so-called syllable-timed language, the duration of an word best correlates with the number of syllables in the word. In a mora-timed language it correlates best with the number of moras. The difference between the two is evident only when there is a difference between the number of syllables in a word and the number of moras -- which means, you have to have long vowels or some reason to posit moraic coda consonants. Neither of these conditions holds in Swahili, so it is impossible to determine if the language is syllable-timed or mora-timed.

However: the claim for there being any "mora-counting" Bantu languages is dubious. The implication of mora-counting is that a 4-mora word has twice the duration of a 2-mora word and half the duration of an 8-mora word. Instead, Bantu languages tend reduce duration of long vowel preceding the penult or antepenult (sometimes resulting in actual phonological shortening). This is more like a stress-timed system (there is at most one stress, on the penult, so the last two syllables tend to be a length unit and the stuff before – regardless of the number of syllables – tends to be a length unit, which can stretchto very many syllables)

There is not very good evidence in Swahili for the mora, since there is no vowel length, and it does not have a mora-counting tone system. The evidence for vowel length in orthographic kondoo, njoo is not strong: vowel doubling in spelling is a way of indicating final stress. The alternative is to say that in some words the stress is final (or antepenult, e.g. barábara). This is in contrast with many (don't know about most) Bantu languages which do provide ample evidence for the concept of mora. There is some confusion out there about the status of "moraicity", since there are surface-contrastive syllabic nasals in Swahili, and these are frequently called "moraic nasals". But that is simply driven by a particular theory of syllables and syllabicity, where "syllabic" i.e. "being a syllable peak" is taken to be defined as "being moraic". If you don't insist on that equation (I don't), then there is no evidence for "moraic nasals". Ngunga and Hyman have a paper illustrating the problem, from Yao (a language with excellent evidence for the concept of mora).

  • 1
    I took a field methods course in 1967, plus or minus, from Terry Langendoen, in which the informant spoke Tumbuka, a Bantu language similar to Swahili. Our informant, Patrick Mkandawire, luckily, was highly educated and completely fluent in English. If there was something the class and Terry didn't understand, we could simply ask Patrick, who would switch into his teaching-Tumbuka mode. The prosody seemed unproblematic, to me (as an English speaker), much like Spanish.
    – Greg Lee
    Commented Feb 28, 2019 at 12:45

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.