Phonotactics

Phonotactics

Korean syllable structure is maximally /CgVC/, where g is a glide /j/ or /w/. Any consonant but /ŋ/ may occur initially, whereas only /p, t, k, m, n, ŋ, l/ may occur finally. Sequences of two consonants may occur between vowels, as outlined above. However, morphemes may also end in CC clusters, which are only both expressed when followed by a vowel. When the morpheme is not suffixed, one of the consonants is not expressed; if there is a /h/, which cannot appear in final position, it will be that; otherwise it will be a coronal consonant, and if the sequence is two coronals, then the voiceless one (/s, tʰ, tɕ/) will drop, and /n/ or /l/ will remain. That is, no sequence reduces to [t̚] in final position.

Sequence	ㄳ gs	ㄺ lg	ㄵ nj	ㄶ nh	ㄽ ls	ㄾ lt	ㅀ lh	ㅄ bs	ㄼ lb	ㄿ lp	ㄻ lm
Medial allophone	[k̚s͈]	[lɡ]	[ndʑ]	[n(ɦ)]	[ls]	[ltʰ]	[l(ɦ)]	[p̚s͈]	[lb]	[lpʰ]	[lm]
Final allophone	[k̚]		[n]		[l]			[p̚]			[m]

When such a sequence is followed by a consonant, the same reduction takes place, but a trace of the lost consonant may remain in its effect on the following consonant. These effects are the same as in a sequence between vowels: an elided obstruent will leave the third consonant fortis, if it's a stop, and an elided //h// will leave it aspirated. Most conceivable combinations do not actually occur;<ref group=note>For example, morpheme-final //lp// only occurs in verb roots such as 밟 balb, and is only ever followed by the consonants d, j, g, n.</ref> a few examples are: //lh-tɕ// = [ltɕʰ], //nh-t// = [ntʰ], //nh-s// = [ns͈], //ltʰ-t// = [lt͈], //ps-k// = [p̚k͈], //ps-tɕ// = [p̚t͈ɕ]; also //ps-n// = [mn], as /s/ has no effect on a following /n/, and //ks-h// = [kʰ], with the /s/ dropping out.

When the second and third consonants are homorganic obstruents, they merge, becoming fortis or aspirate, and—depending on the word, and a preceding //l// might not elide: //lk-k// is [lk͈].

An elided //l// has no effect: //lk-t// = [k̚t͈], //lk-tɕ// = [k̚t͈ɕ], //lk-s// = [k̚s͈], //lk-n// = [ŋn], //lm-t// = [md], //lp-k// = [p̚k͈], //lp-t// = [p̚t͈], //lp-tɕ// = [p̚t͈ɕ], //lpʰ-t// = [p̚t͈], //lpʰ-tɕ// = [p̚t͈ɕ], //lp-n// = [mn].

Among vowels, the sequences /*jø, *jɯ, *ji, *wo, *wɯ, *wu/ do not occur, and it is not possible to write them using standard hangul.<ref group="note">While 워 is romanized as wo, it does not represent /wo/, but rather /wʌ/ instead.</ref> The semivowel [ɰ] only occurs in the diphthong /ɰ͡i/.