|Native to||Okinawa, Japan|
|(mostly over age 20 cited 1989)|
The Miyakoan language (/ My?kufutsu/My?kufuts? [mja:kufutss?] or /? Sumafutsu/S?mafuts?) is a diverse dialect cluster spoken in the Miyako Islands, located southwest of Okinawa. The combined population of the islands is about 52,000 (as of 2011). Miyakoan is a Southern Ryukyuan language, most closely related to Yaeyama. The number of competent native speakers is not known; as a consequence of Japanese language policy which refers to the language as the Miyako dialect (?, Miyako h?gen), reflected in the education system, people below the age of 60[timeframe?] tend to not use the language except in songs and rituals, and the younger generation mostly uses Japanese as their first language. Miyakoan is notable among the Japonic languages in that it allows non-nasal syllable-final consonants, something not found in most Japonic languages.
The most divergent variant is that of Tarama Island, the farthest island away. The other variants cluster as Ikema-Irabu and Central Miyako. Given the low degree of mutual intelligibility, Tarama language is sometimes considered a distinct language in its own right.
An illustrative lexeme is the name of the plant Alocasia (evidently an Austronesian loan: Tagalog /bi:?a?/). This varies as Central Miyako (Hirara, ?gami) /bi?kas:a/, Ikema /b?b?:?am:/, Irabu (Nagahama) /bb?:?as:a/, Tarama /biv:uas:a/.
The description here is mostly based on the ?gami variant, the Central Miyakoan variant of the smallest of the Miyako islands, from Pellard (2009). There is additional description based on the Irabu variant, the Ikema-Irabu variant of the second largest of the Irabu islands. 
Central Miyakoan variants do not have pitch accent; therefore, they are of ikkei type. Tarama distinguishes accent on the phonological word (stem plus clitics), e.g. /juda?mai neen/, /jadu?mai? neen/, /madumai neen/,
There are five vowels in ?gami.
/?/ is truly unrounded, unlike the compressed Japanese u. It is centralized after /s/. /u/ is rounded normally, but varies as [?]. /?/ varies from [e] to [æ].
Numerous vowel sequences occur, and long vowels are treated as sequences of identical vowels, keeping the inventory at five.
Historical *i and *u centralized and merged to /?/ as *e and *o rose to /i/ and /u/. The blade of the tongue in /?/ is close to the alveolar ridge, and this feature has been inaccurately described as "apical" (it is actually laminal). In certain environments /?/ rises beyond vowel space to syllabic /s?/ after /p/ and /k/ (especially before another voiced consonant) and, in variants that have voiced stops, to /z?/ after /b/ and /?/:
?gami vowels other than /?/ are not subject to devoicing next to unvoiced consonants the way Japanese high vowels are. Sequences of phonetic consonants have been analyzed by Pellard (2009) as being phonemically consonantal as well.
In Irabu there are five main vowels and two rare mid vowels that occur in loanwords and some clitics. 
In ?gami there are nine consonants, without a voicing contrast. (Most Miyakoan variants do distinguish voicing.)
The plosives tend to be somewhat aspirated initially and voiced medially. There are maybe a dozen words with optionally voiced initial consonants, such as babe ~ pape (a sp. of fish) and gakspstu ~ kakspstu 'glutton', but Pellard suggests they may be loans (babe is found in other variants, and gaks- is a Chinese loan; only a single word gama ~ kama 'grotto, cave' is not an apparent loan).
/k/ may be spirantized before /?/: kaina 'arm' [k?in? ~ x?in?], a?ka 'I (nominative)' [?k? ~ ?x? ~ ].
/n/ is [?] at the end of a word, and assimilates to succeeding consonants ([m~n~?]) before another consonant. When final [?] geminates, it becomes [nn]; compare tin [ti?] 'silver' with tinnu [tinnu] 'silver (accusative)'. It tends to devoice after /s/ and /f/. /m/, on the other hand, does not assimilate and appears finally unchanged, as in mku 'right', mta 'earth', and im 'sea'.
/f/ is labiodental, not bilabial, and /s/ palatalizes to [?] before the front vowels /i ?/: pssi [pi] 'cold'. Some speakers insert an epenthetic [t] between /n/ and /s/ in what would otherwise be a sequence thereof, as in ansi [?n?i ~ ?nt?i] 'thus'.
/?/ is clearly labiodental as well and tends to become a fricative [v] when emphasized or when geminated, as in /ku/ [kuvv?] 'calf'. It can be syllabic, as can all sonorants in ?gami: vv [v?:] 'to sell'. Final /?/ contrasts with the high back vowels: /pa?/ 'snake', /pau/ 'stick', /pa?/ 'fly' are accusative [p?vvu, p?uju, pu] with the clitic -u.
Various sequences of consonants occur (mna 'shell', sta 'under', fta 'lid'), and long consonants are bimoraic (sta [s?.t?] fta [f?.t?], pstu [ps?.tu]), so they are analyzed as consonant sequences as well. These can be typologically unusual:
Geminate plosives do not occur, apart from a single morpheme, the quotative particle tta.
There are a few words with no voiced sounds at all (compare Nuxálk language § Syllables):
The contrast between a voiceless syllable and a voiced vowel between voiceless consonants can be seen in kff puskam [k?f?:pusk?m] 'I want to make (it)', ff?nk?i [f?:i] 'to?the.comb', and paks?nu?tu [p?ksn?udu] 'bee?NOM?FOC' (with a devoiced nasal after s). There is a contrast between ff?m?i 'comb?INCL' and ffu?m?i 'shit?INCL'. With tongue twisters, speakers do not insert schwas or other voiced sounds to aid in pronunciation:
The minimal word is either VV, VC, or CC (consisting of a single geminate), as in aa 'millet', ui 'over', is 'rock', ff 'comb'. There are no V or CV words; however, CCV and CVV words are found, as shown above.
Syllabification is difficult to analyze, especially in words such as usnkai (us-nkai) 'cow-DIR' and saiafn (saiaf-n) 'carpenter-DAT'.
There are 15-16 consonants in Irabu, which do have a voicing contrast.