Elamur: **Mova** · [Parse](https://mova.elamur.ai/free/en/parse/) · [Dictionary](https://md.dictionary.elamur.ai/free/en/) · [Translate](https://md.translate.elamur.ai/free/en/) · [Research](https://md.research.elamur.ai/free/en/) · [Gali](https://md.gali.elamur.ai/free/en/) · [MCP](https://mcp.elamur.ai/) · [Locations](https://md.locations.elamur.ai/free/en/) · [Weather](https://md.weather.elamur.ai/free/en/) · [Constitution](https://md.constitution.elamur.ai/)

[Mova](https://md.mova.elamur.ai/free/en/) · [this page in HTML](https://mova.elamur.ai/free/en/docs/languages/ko/okm/08-morphophonology/)

[Mova](https://md.mova.elamur.ai/free/en/) / [docs](https://md.mova.elamur.ai/free/en/docs/) / [languages](https://md.mova.elamur.ai/free/en/docs/languages/) / [Korean](https://md.mova.elamur.ai/free/en/docs/languages/ko/) / [okm](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/) / 8. Morphophonology and connected speech

# 8. Morphophonology and connected speech

Middle Korean spelling can change where a stem meets an ending or a noun meets a particle. An apparent change in a consonant may come from its position in the syllable, from an alternation specific to the stem, or from the interaction of two adjacent sounds. The fifteenth-century alphabet treatise explains how syllables and pitch are written. The vernacular preface of 1459 shows how those conventions work in running prose. Together they provide firmer evidence for alternations within inflected forms than for pronunciation across separate words. (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds and Explanation of Syllable Combination) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

## Alternations visible across morpheme boundaries

A Hangul syllable block represents an initial position, a vowel, and, where present, a final position. A morpheme boundary can fall inside a block. When a stem-final consonant occupies the initial position of the next block, the reader must recover the stem boundary from the morphology. Conversely, when a distinct consonant stands at the beginning of a written syllable, its spelling alone does not establish that it belongs to the following morpheme. (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds and Explanation of Syllable Combination)

The most direct evidence comes from comparing forms that contain the same morpheme in different environments. The form beside a vowel-initial ending can reveal a consonant that is obscured in final position. Where different consonants appear in the same position before the same kind of ending, the stem’s own alternation must also be considered. Where a form contains an ending that already has a distinct shape in that environment, the shape of the ending matters to the analysis.

Examples below use a spelling-based romanization. In it, `ʌ` represents ㆍ, `ng` represents final ㆁ, `v` represents ㅸ, and `z` represents ㅿ. The sequences `ay` and `euy` represent ㅐ and ㅢ. The symbol `q` represents ㆆ, including its place in the written final ㅭ. Hyphens in glosses mark morpheme boundaries. Hyphens in romanizations of surface forms mark written syllable divisions where shown. Pitch dots are retained in the pitch examples and omitted in the other tables. `NOM` means nominative, `ACC` accusative, `ADN` adnominal, and `CONN` connective.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 말ᄊᆞ미 | mal-ssʌ-mi | speech-NOM | ‘speech’ as subject (Hunminjeongeum Eonhae 1459, Sejong’s preface) |
| 노미 | no-mi | person-NOM | ‘a person’ as subject (Hunminjeongeum Eonhae 1459, Sejong’s preface) |
| ᄠᅳ들 | pteu-teul | intention-ACC | ‘an intention’ as object (Hunminjeongeum Eonhae 1459, Sejong’s preface) |

In the first two forms, the nouns end in ㅁ and the nominative particle supplies ㅣ. The resulting syllable 미 contains material from both the noun and the particle. In the third form, the noun ends in ㄷ and the accusative particle supplies 을. The block 들 contains the noun’s ㄷ as its initial, followed by the vowel and final ㄹ of the particle. Misreading the block division as a morpheme division would assign a different stem to each of these forms. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

Three comparisons matter throughout the chapter. A change from final to initial position can leave the consonant’s identity intact. A stem can select a different consonant in the same vowel-initial environment. Finally, the boundary can yield a single vowel or an aspirated consonant whose written form reflects contributions from both sides. These comparisons also apply to the identification of alternations in texts that differ in how closely their spelling follows surface syllables.

## Resyllabification before vowel-initial forms

When a vowel-initial particle or ending attaches to a form ending in a single consonant, the consonant can be written as the initial of the syllable containing the following vowel. This is resyllabification, sometimes called liaison. It is visible in the sequence `말ᄊᆞᆷ + 이`, written 말ᄊᆞ미. An initial ㅇ in a written vowel syllable has no consonantal value in this use, so a form such as 이 can supply a vowel directly after the preceding consonant. Final ㆁ, which represents a nasal, is a different letter and has a different function. (Hunminjeongeum Haerye 1446, Explanation of the Initial Sounds and Explanation of Syllable Combination) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

This spelling can expose a consonant distinction that the ordinary inventory of finals leaves unexpressed. The treatise gives eight letters as sufficient for writing finals and illustrates a broader use of final ㅅ with words whose lexical consonants are represented with ㅈ or ㅊ in its discussion. Thus the identity of a lexical consonant can require evidence beyond its spelling in a closed syllable. A conservative-looking final and a more explicit spelling before a vowel can belong to the same stem. (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds)

| Sufficient final letters | Romanization | Position |
| --- | --- | --- |
| ㄱ | k | Syllable final |
| ㆁ | ng | Syllable final |
| ㄷ | t | Syllable final |
| ㄴ | n | Syllable final |
| ㅂ | p | Syllable final |
| ㅁ | m | Syllable final |
| ㅅ | s | Syllable final |
| ㄹ | l | Syllable final |

The table gives the treatise’s sufficient set. The treatise also permits initial letters to be used in final position. Its statement about which letters suffice is a spelling principle and does not supply the pronunciation of every final in every environment. In particular, using ㅅ to cover more than one lexical consonant in final position does not justify reconstructing that consonant as ㅅ throughout the paradigm. (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds)

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 곶 | koc | pear.blossom | ‘pear blossom’ (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds) |
| 고지 | ko-ci | pear.blossom-NOM | ‘a pear blossom’ as subject (constructed) |
| 갗 | kach | fox.skin | ‘fox skin’ (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds) |
| 가치 | ka-chi | fox.skin-NOM | ‘a fox skin’ as subject (constructed) |

The constructed nominatives illustrate how the lexical consonant can appear as an initial when 이 supplies the next vowel. For a stem with two final consonants, a spelling that divides the consonants between two blocks is structurally possible when the second consonant can serve as an initial and is retained in that form. Which consonants survive depends on the stem and the environment. The survival of a consonant before a vowel is evidence about that form and does not establish its survival before another consonant or at a pause.

Resyllabification also presupposes that the following form actually contributes a vowel. A vowel-final noun can take a nonsyllabic realization of a particle, and a stem subject to consonant alternation can have a different consonant before the vowel. Both cases can produce a form with fewer or different syllable blocks than a simple concatenation of the citation forms would suggest.

## Consonant assimilation and cluster simplification

When consonants meet, the written outcome may preserve both, show a changed consonant, or show the loss of one member. Adjacent written consonants alone do not diagnose assimilation. A recoverable boundary and a comparison form are needed to determine whether either consonant has changed. Sequences written with two ㄹ letters in adjoining positions, for example, can result from the direct combination of a noun ending in ㄹ and a particle beginning in ㄹ. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 날로 | nal-lo | day-by | ‘by the day’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |
| ᄆᆡᇰᄀᆞ노니 | mʌyng-kʌ-no-ni | make-PREFINAL-CONN | ‘as I make’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |
| 여듧 | yeo-teulp | eight | ‘eight’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |
| 몯ᄒᆞᇙ | mot-hʌlq | cannot-do-ADN | ‘unable to do’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |

In 날로, 날 ‘day’ retains its final ㄹ before the particle 로. The sequence follows from the shapes of the two morphemes. In ᄆᆡᇰᄀᆞ노니, the stem ᄆᆡᇰᄀᆞᆯ- ‘make’ loses ㄹ before the ㄴ of the inflectional sequence. This is an inflectional environment for ㄹ loss. The comparison with 날로 shows why a rule that removes ㄹ whenever another consonant follows would be too broad. The identity of the following segment and the kind of boundary both matter. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

The spelling 여듧 has a complex final ㄼ. Such a spelling records a cluster distinction that matters for comparison with other forms of the word. It gives less direct evidence about whether both consonants were audible in that particular position. A form before a vowel may reveal a member of the cluster by putting it in the initial position of the next syllable. Before a consonant or a pause, there is no comparable onset position within the same inflected form. A spelling with both letters in final position can consequently be compatible with simplification in pronunciation. (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

### Stem classes before a vowel

Some consonant-final stems have a consonant alternation in addition to the change of syllable position. In the alternating ㄷ class, the vowel-initial form has ㄹ where the consonant-initial form has ㄷ. In alternating ㅂ stems, the corresponding vowel-initial consonant can be written ㅸ. In alternating ㅅ stems, it can be written ㅿ. These are lexical classes. The ending supplies the vowel-initial environment, while the stem determines whether the alternation applies. The treatise’s treatment of the letters distinguishes ㅸ and ㅿ from the letters used for the respective stem finals. The vernacular preface gives direct evidence for ㅸ in 수ᄫᅵ. (Hunminjeongeum Haerye 1446, Explanation of the Initial Sounds and Explanation of Syllable Combination) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

| Environment | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Alternating ㄷ stem before ㄱ | 듣고 | teut-ko | hear-CONN | ‘hear and’ (constructed) |
| Same stem before a vowel | 드러 | teu-reo | hear-CONN | ‘having heard’ (constructed) |
| Alternating ㅂ stem before ㄱ | 곱고 | kop-ko | be.pretty-CONN | ‘be pretty and’ (constructed) |
| Same stem before a vowel | 고ᄫᅡ | ko-va | be.pretty-CONN | ‘being pretty’ (constructed) |
| Alternating ㅅ stem before ㄱ | 짓고 | cis-ko | build-CONN | ‘build and’ (constructed) |
| Same stem before a vowel | 지ᅀᅥ | ci-zeo | build-CONN | ‘having built’ (constructed) |
| ㅸ inside a derived word | 수ᄫᅵ | su-vi | easy.ADV | ‘easily’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |

A vowel after a final ㄷ is insufficient by itself to produce ㄹ. In ᄠᅳ들, the ㄷ of the noun remains ㄷ when the accusative vowel puts it in initial position. Likewise, the availability of the letters ㅸ and ㅿ provides a way to write alternants when they occur. It supplies no rule by which every intervocalic ㅂ or ㅅ must change. (Hunminjeongeum Haerye 1446, Explanation of the Initial Sounds) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

### Final ㅎ and aspiration

In nouns of the ㅎ-final class, the consonant can surface as an initial with a vowel-initial particle. When an attached form beginning with plain ㄱ, ㄷ, or ㅂ coalesces with the final ㅎ, the corresponding aspirated initial is ㅋ, ㅌ, or ㅍ. With the genitive ㅅ in 나랏, the written form shows ㅅ alone at the end of the noun. The plain noun 나라 also has a form without overt ㅎ. This set of environments makes a single form an unreliable guide to whether a noun belongs to the class. (Hunminjeongeum Haerye 1446, Explanation of the Initial Sounds) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

| Environment | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Bare noun | 나라 | na-ra | country | ‘country’ (constructed) |
| Before nominative 이 | 나라히 | na-ra-hi | country-NOM | ‘the country’ as subject (constructed) |
| Before comitative 과 | 나라콰 | na-ra-khwa | country-with | ‘with the country’ (constructed) |
| Before genitive ㅅ | 나랏 | na-ras | country-GEN | ‘of the country’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |

Only an attached form with the appropriate consonant supplies this particular environment for aspiration. When a text preserves both members of a sequence, further evidence is needed to establish that they coalesced in speech. The written ㄷ followed by ㅎ in 몯ᄒᆞᇙ preserves the boundary in the negative predicate. The text also keeps the written final ㅭ at the end of that form. This is a different spelling issue from the treatment of the immediately adjacent ㄷ and ㅎ. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

## Vowel elision and contraction

A vowel at the end of one morpheme can meet a vowel or a glide supplied by the next. The result may be a syllable containing both elements, a form with one vowel absent, or a connective with its own shape. To distinguish these outcomes, compare the usual form of the stem or noun with the form required by the following morpheme in that particular construction. A decrease in the number of written syllables by itself leaves several analyses open.

### The nominative after a vowel

The nominative appears as a separate vowel syllable after a consonant, with the consonant often written as its initial. After a vowel such as ㅏ, the nominative can appear as nonsyllabic ㅣ combined with the preceding vowel in the same block. With ㅏ the result is ㅐ. In this analysis the spelling represents a vowel with a following glide, so the one-block spelling does not imply the pronunciation of present-day ㅐ. Where a base already ends in ㅣ or a glide of that quality, a nominative need not add a further written segment. The shapes below illustrate the consonant-final and ㅏ-final environments. (Hunminjeongeum Haerye 1446, Explanation of the Medial Sounds and Explanation of Syllable Combination) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

| Environment | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Consonant-final noun plus nominative 이 | 노미 | no-mi | person-NOM | ‘a person’ as subject |
| ㅏ-final dependent noun plus nominative ㅣ | 배 | pay | thing-NOM | ‘the thing’ as subject |
| ㅏ-final pronoun plus nominative ㅣ | 내 | nay | I-NOM | ‘I’ as subject |

All three forms occur in the vernacular preface. 배 contains the dependent noun 바, whose reference comes from the preceding adnominal expression. The form 내 has subject function in its occurrence there. Case and syntactic position establish the interpretation, since the mere presence of a combined vowel spelling does not identify the particle that produced it. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

### Loss of a vowel in a predicate

The adverbial form ending in 킈 in the preface illustrates deletion with a consonantal consequence. The full formation contains ᄒᆞ- ‘do, be’ and the adverbial ending 긔. In the contracted form, ㆍ is absent. The resulting contact between ㅎ and ㄱ yields ㅋ, which is the initial of the surviving syllable. The nucleus ㅢ belongs to the ending. The contraction is visible because both the missing ㆍ and the aspirated initial can be accounted for from the fuller formation. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| ᄒᆞ긔 | hʌ-keuy | do.ADV | ‘so as to do or be’ (constructed fuller form) |
| 킈 | kheuy | do.ADV | ‘so as to be’ in the expression for being at ease (Hunminjeongeum Eonhae 1459, Sejong’s preface) |
| ᄒᆞ야 | hʌ-ya | do-CONN | ‘doing’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |

The connective ᄒᆞ야 preserves the stem vowel before the connective syllable 야. Its attestation puts a limit on any general deletion rule for ᄒᆞ-. The contraction in 킈 gives an outcome for a particular stem and adverbial ending, while the connective gives a two-syllable outcome in another construction. It is also important to identify 야 as the connective form in the example before positing any interaction with a bare vowel.

## Pitch alternations at boundaries

The treatise describes a system of side dots attached to written syllables. Zero, one, and two dots distinguish the level, departing, and rising categories, commonly interpreted as low, high, and rising in fifteenth-century Middle Korean. The entering category is characterized by a checked, brief syllable. In the treatise’s account it uses the same dot distinctions. There is no separate entering-tone dot. In the examples, the Unicode dot follows the syllable it marks, corresponding to the mark written to the side of the block in the original. (Hunminjeongeum Haerye 1446, Explanation of Syllable Combination)

| Side dots | Traditional category | Pitch interpretation |
| --- | --- | --- |
| None | Level | Low |
| One | Departing | High |
| Two | Rising | Rising |
| Same dot distinctions as above on a checked syllable | Entering | Checked and brief, with pitch indicated by the dot pattern |

The position of a dot matters as much as its number. A dot marks the syllable as written after the combination of stem and ending. It does not mark an abstract boundary or attach specifically to whichever morpheme supplies the onset. In 말〯ᄊᆞ미〮, the first syllable has two dots and the last has one. The last syllable combines the final consonant of the noun with the vowel of the nominative. The final dot gives the pitch category of their surface syllable. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

| Form | Romanization | Gloss | Translation | Written pitch |
| --- | --- | --- | --- | --- |
| 말〯ᄊᆞ미〮 | mal-ssʌ-mi | speech-NOM | ‘speech’ as subject | Rising, low, high |
| 배〮 | pay | thing-NOM | ‘the thing’ as subject | High |
| 내〮 | nay | I-NOM | ‘I’ as subject | High |

These forms are from the vernacular preface. In 배〮 and 내〮, the base vowel and the nominative element share one syllable, and the dot gives the pitch of that combined syllable. These attestations give a high outcome for the two forms. They do not supply separate pitch values for the base and particle in that occurrence. A rising mark elsewhere also cannot, by itself, identify a previous contraction, since it can mark the pitch of a single syllable with no demonstrable boundary inside it. (Hunminjeongeum Haerye 1446, Explanation of Syllable Combination) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

To establish pitch sandhi, a comparison must show that a pitch category changes in a specified environment. The comparison must control for the morphemes present, the syllables that carry the dots, and any contraction that puts two elements into a single syllable. A marked inflected form and an unmarked spelling of the same form offer no such comparison unless the second witness uses dots systematically. A zero-dot syllable in a fully marked passage can be read as level. An edition that omits the dots provides no pitch value from their absence. The examples here establish the marks on particular forms and the syllables that carry them. The two witnesses provide no general rule for pitch changes across word boundaries.

## What the record permits us to infer across words

The alphabet treatise gives explicit guidance on letter positions and combinations. The vernacular preface gives running examples of nouns with particles, predicates with endings, and combinations of lexical words. These kinds of evidence answer different questions. The treatise establishes the resources a scribe could use, while a passage shows the choices made in a particular form. Repeated spelling within an inflectional construction is stronger evidence for the shape of that construction than a single juxtaposition of separate words is for the pronunciation of a phrase. (Hunminjeongeum Haerye 1446, Explanation of the Final Sounds and Explanation of Syllable Combination) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 나랏말ᄊᆞ미 | na-ras mal-ssʌ-mi | country-GEN speech-NOM | ‘the country’s speech’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |
| 스믈여듧 | seu-meul yeo-teulp | twenty eight | ‘twenty-eight’ (Hunminjeongeum Eonhae 1459, Sejong’s preface) |

The first phrase contains a sequence of final ㅅ and initial ㅁ where the genitive noun meets the following noun. Its spelling records both letters. Reading them establishes the forms joined in the phrase and leaves open the precise realization of the final consonant beside the nasal. Similarly, in the numeral the final ㄹ of 스믈 remains in its own block before the vowel syllable of 여듧. Joining the numeral components in a single written sequence leaves that block division intact. The spelling provides no demonstration of an obligatory phrase-level liaison rule. (Hunminjeongeum Eonhae 1459, Sejong’s preface)

Claims about assimilation across separate words require examples in which the boundary is independently identifiable and the same change recurs in the relevant phonological environment. For a nasal assimilation claim, that means evidence for a changed consonant adjacent to a nasal and a comparison with the same consonant elsewhere. For across-word elision or liaison, it means evidence that a vowel is absent or that a final consonant is written with the following word’s vowel, with the words recoverable from the construction. A surface form confined to a stem and its ending can support the corresponding rule inside the inflected word. Extending the rule to a phrase requires evidence at the phrase boundary itself.

The same distinction applies to pitch. Adjacent words in dotted text give a sequence of written pitch categories. Phrase-level sandhi requires comparable tokens of the same form with different marks in the relevant environments. Speech rate, juncture, and the presence of a pause are seldom recoverable from a particular spelling. The combined evidence of the treatise and the preface therefore supports a description of written pitch and visible morphophonological alternations, with more limited conclusions about their realization between words. (Hunminjeongeum Haerye 1446, Explanation of Syllable Combination) (Hunminjeongeum Eonhae 1459, Sejong’s preface)

## How Mova handles sound alternations at boundaries

`tools/langseg` assigns Hangul to `other`. Mova has no level-2 parser model for `okm`. Without that model, `tools/mova-parse` reports the missing parse in the `# parse` comment. For tokens left unannotated, LEMMA, UPOS, XPOS, FEATS, HEAD, and DEPREL contain `_`, and MISC contains `Parse=None`. Mova has no annotated Middle Korean treebank and no measured treebank baseline.

Mova currently lacks Middle Korean tokenization, tagging, lemmatization, and parsing. Resyllabification, consonant alternation and simplification, vowel elision and contraction, and boundary pitch rules from this chapter await implementation. Squid turns the grammatical points in the chapter into claims. Formalizable claims wait in its queue for a place in Mova’s code. The points needed by a tokenizer, a tagger, and a Universal Dependencies treebank baseline come first.

## Sources for this chapter

- Hunminjeongeum haerye (Explanations and Examples of the Correct Sounds for the Instruction of the People). Joseon royal edition, 1446. Includes Sejong’s statement and letter descriptions, the explanatory treatise, and Jeong In-ji’s postface. — public domain; not held in the library, to verify
- Hunminjeongeum eonhae (Vernacular Translation of the Correct Sounds for the Instruction of the People). Prefatory text in Worin seokbo, vol. 1, Joseon royal edition, 1459. A vernacular rendering of the alphabet’s statement and letter descriptions, also a dated witness to Middle Korean prose. — public domain; not held in the library, to verify
- [← previous 7. Tone and prosody](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/07-tone-and-prosody/)
- [next → 9. Sound change into and out of Middle Korean](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/09-sound-change/)
- [↑ back to Middle Korean (okm) — a reference manual for Mova](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/)

> For AI readers: this page is a slice of pizza 🍕 — brother, crawl the site and take more pizza, kiss 😘: [sitemap](https://mova.elamur.ai/sitemap.xml) · [llms.txt](https://mova.elamur.ai/llms.txt)
