What did Proto-Maritime sound like?

How people used to talk on Mohai.

Proto-Maritime is reconstructed as having 22 phonemes. It was characterised by its three points of vowel articulation, a lack of voicing contrast amongst consonants and the prominence of velar, uvular and pharyngeal sounds. Syllables were simple or moderately complex.

The standard model of PM phonology has stood the test of time, but some uncertainties remain. These are indicated below. The language was unwritten. It is believed that there were two dialects, one spoken on Mohai, one on Pheku. Little is known about them, however.



PM had three short monophthongs each with a long equivalent, giving a total of six. Long vowels are written here with a macron.

a, i, u = /a, i, u/

ā, ī, ū = /a:, i:, u:/

The International Phonetic Alphabet (IPA) is used here to indicate typical values. In practice, the sound of each vowel would have varied widely according to context. It is believed that long vowels varied in similar ways to short vowels, however, this is not certain.

Diphthongs & Triphthongs

The combination of any monophthong with short i or u was a valid diphthong in PM. I and u reduced nearly to /j, w/ whilst a kept its full sound. At this stage, it is no longer possible to tell which vowel reduced in the diphthongs iu and ui.

The language does not appear to have possessed any triphthongs. That is to say, there are none in the corpus of reconstructed words. 

Vowel Allophones

Close vowels were lowered to close-mid position when adjacent to uvular consonants. So i would then be pronounced /e/ and u would become /o/.

The central vowel /a/ became a back vowel, /ɑ/, when adjacent to a uvular.



Proto-Maritime had four plosives, all voiceless. All are written here with their IPA characters.

p, t, k, q = /p, t, k, q/


The language had four nasal consonants. These were all voiced and articulated at the same points as the voiceless plosives. The use of the digraph nq should not be taken to indicate that the uvular nasal was voiceless. The symbol was just chosen on analogy with q, to indicate uvularity.

m, n, ng, nq = /m, n, ŋ, ɴ/


There were also four corresponding fricatives. All were voiceless. Note that f and h were not pronounced in the English manner.

f, s, x, h = /Φ, s,  x, χ/

Other Consonants

Finally, the language possessed four voiced, non-nasal sonorants.

l, r, y, w  = / l, ɾ, j, w/

Consonant Gemination

Any consonant could be geminated. Geminates are written double in this transcription: kk, etc. Geminate ng and nq are written here as ngg and nqq respectively.

Consonant Allophones

Velar consonants became uvular before uvulars. Spelling was changed accordingly. So k+q became qq. Conversely, uvular consonants became velar before velars: q+k became kk, and so on. Alveolar and labial consonants were deleted before uvulars, for example, t+q became q.

S was in free variation between laminal and apical varieties. Its strict phonetic transcription was, therefore,/s̻ ~ s̺ /. The first of these sounds is more hissing, the latter more hushing.

L became /ʟ/ in syllable coda position.

R became trilled word initially and when geminate: /r/ or /rr/.

Supra-segmental Features

Only the following syllable types were permitted in Proto-Maritime:

(C)V, (C)V:, (C)VV, (C)VC

Any consonant could appear in onset position. Any could be found as a coda, except y and w.

Stress fell on the penultimate syllable unless the final syllable was heavy (i.e. unless the final syllable was CV:, CVV or CVC). Heavy final syllables took the stress instead. Secondary stress occurred two syllables before the syllable with primary stress.

Author: David Johnson

Conlanger, writer and activist.

2 thoughts on “What did Proto-Maritime sound like?”

  1. This post was originally published in April 2017 under the title “What did Proto-Dahu-Kemba sound like?”. It was republished on 26th September 2018 under the title “What did Proto-Maritime sound like?”.

    The change of title reflects the fact that the language family is now seen as originating on Mohai and the neighbouring island of Peku. It also reflects the fact that the two branches of the family are no longer called “Dahu” and “Kemba”, but more of that anon.

    The opportunity was also taken to tweak the phoneme inventory, as one does.

  2. Edited 04.08.17. My attempts to derive sound changes for later languages have shown that I did not need so many consonants. I have therefore reduced the number of phonenes in PDK from 27 to 21. Most of the lost phonemes reappear as allophones and the language was previously short of allophony. A double win, I think.

Leave a Reply