Mova0.6 pre-release

Okinawan: the flag of Okinawa Prefecture, a white ring in a red disc on white 16. Derivation, compounding and reduplication

Central Okinawan forms words by attaching affixes, combining bases and repeating material from a base. It also incorporates loans and preserves formations whose meanings have become conventional. This chapter uses Central Okinawan forms associated with Shuri and Naha. Hyphens in the examples show the analysis of a form, and doubled vowel letters indicate length.

Productive prefixes and suffixes

A derivational affix contributes to the meaning or word class of a base. Some affixes give the speaker a way to form a new word on a familiar model. Others recur in a group of established words whose membership has to be learned. The range of bases an affix accepts and the meaning it contributes are separate parts of its description. A diminutive suffix, for example, can be available on many nouns while giving an affectionate reading only in certain contexts. Chamberlain's description records older Central Okinawan usage. Forms documented elsewhere in the Ryukyus belong to the grammar of the variety where they are used (Chamberlain 1895, Grammar; Shimoji and Pellard 2010, Introduction).

AffixBase and outputContributionForm, romanization, gloss and translation
u-Before nouns in established honorific vocabularyRespectful or polite referenceうちゃ u-cha HON-tea ‘tea’ in polite usage (constructed)
gu-Before nouns in established honorific vocabulary, including words of Sino-Japanese originRespectful or polite reference, also found in fixed expressionsぐぶりー gu-burii HON-discourtesy ‘a discourtesy’, used in expressions of apology or leave-taking (constructed)
-gwaaAfter a noun referring to a person, animal or thingSmallness, affection or familiarity, depending on the noun and the situationわらびぐゎー warabi-gwaa child-DIM ‘little child’ (constructed)
-nchuAfter a place name or a base identifying a communityPerson associated with that place or communityうちなーんちゅ uchinaa-nchu Okinawa-person ‘Okinawan person’ (constructed)

The respectful prefixes are sensitive to lexical convention. The origin of the noun can help explain which prefix appears, but a speaker learns the combination with the noun or expression in which it occurs. In gu-burii, for example, the polite force is part of the expression in which the noun is used. The choice of prefix belongs to the form of the word. The choice to use that word belongs to the interaction.

With -gwaa, physical size is one possible contribution. On nouns for people and animals, the speaker's attitude can carry more of the meaning. The base in warabi-gwaa already identifies a child. The suffix can convey endearment as well as smallness. On a noun for an object, the immediate situation may make relative size the salient interpretation. A gloss such as DIM identifies the formation across these readings.

The person formative -nchu contributes a new referent. uchinaa denotes a place, and uchinaa-nchu denotes a person identified with it. The same pattern is visible in しまんちゅ shima-nchu island-person ‘islander’ (constructed). Here shima can supply the relevant island or local community in context. Established names of groups have conventional forms, so the pattern alone establishes neither the preferred name for every community nor the exact shape of every place-name base.

Position of the formative

The productive affix positions illustrated here are before and after the base. An infix would have to occur inside a base whose parts remain identifiable on either side of it. Sounds that appear or change where two parts meet are evidence about the boundary between those parts. For example, the voiced beginning of the second member of uchinaa-guchi belongs to the shape of that member in the compound.

Central Okinawan has no general productive root-and-pattern system of the kind in which an independently identifiable consonantal root is combined with a vocalic template to derive whole classes of words. The alternations involved in choosing a verb stem belong to conjugation. Affixation uses a base that may itself have alternants, and an expressive word may repeat only part of its material. Both can give a surface form whose base is easier to recognize by comparison with related words.

Nominalization and changes of word class

Nominalization makes a form available for use as a noun or gives a larger expression a noun-like distribution. Word formation concerns the first of these when a base gives rise to a lexical noun. The second concerns the syntax of phrases and clauses. A quality noun can occur with noun-phrase particles and refer to a quality or its degree. A nominalized clause can refer to a fact, event or participant while retaining verbal structure inside the clause.

Quality nouns

A quality noun in -sa names a property or its degree. Its base is the property stem associated with the adjective. In the corresponding predicative form, the material following the stem is -san. The relation is clearest when the forms are compared as a set (Chamberlain 1895, Grammar).

Property basePredicative form, romanization, gloss and translationQuality noun, romanization, gloss and translation
taka- ‘high’たかさん taka-san high-ADJ.PRED ‘is high’ (constructed)たかさ taka-sa high-NMLZ ‘height’ (constructed)
chura- ‘beautiful, clean’ちゅらさん chura-san beautiful-ADJ.PRED ‘is beautiful’ (constructed)ちゅらさ chura-sa beautiful-NMLZ ‘beauty’ (constructed)

With taka-sa, the context may identify the height of a particular object, a degree on a scale, or height as the property under discussion. The noun still names the same property in each use. Its interpretation in a particular phrase comes from the expression it occurs in. For chura-sa, the range of readings follows the senses of the property base and what it describes.

A final sequence resembling -sa supplies a possible segmentation. A corresponding property base and a recurrent relation in meaning support that segmentation. This matters when comparing a transparent quality noun with an established noun whose final syllable happens to have the same shape.

Action nouns and noun-based predicates

A verb stem can occur as a member of a lexical noun. In that use it may name an act, a practice or the product of the act. The utterance noun in the following example contains the stem associated with ‘say’. The whole expression has the distribution of a noun.

FormRomanizationGlossTranslation
むぬいいmunu-iithing-saying‘speaking, manner of speaking’ (constructed)

A stem in a compound may have the same segmental shape as a form used elsewhere in the verb paradigm. Its position and the distribution of the whole construction determine the analysis. The noun munu-ii can name speaking as an activity or a way of speaking. A finite verb of saying heads a clause and can take the dependents of that verb.

A noun that names an activity can also occur with sun ‘do’. The verb carries the verbal inflection, and the noun identifies the activity. This construction is available to established activity nouns and to borrowed vocabulary used as activity nouns.

FormRomanizationGlossTranslation
ゆんたくyuntakuchat.N‘a chat, conversation’ (constructed)
ゆんたく すんyuntaku sunchat do.NPST‘chat, have a chat’ (constructed)

Conversion is a change in lexical class with no overt derivational material. It requires evidence that the same form is established with the distribution of both classes. In yuntaku sun, the presence of sun accounts for the verbal behavior. A noun serving as the first member of a compound likewise has a special distribution within that compound. Such a use on its own establishes the class of the output only. The spelling of an expression as one unit or as two supplies no test for conversion.

Noun and verb compounds

A compound combines bases into a lexical unit. The head usually gives the compound its word class and its broad meaning. In a noun compound with the head on the right, the first member narrows the interpretation of the second. The relation between the members can be one of location, domain, intended use or another association. The lexical meaning comes from the particular combination.

Noun compounds

FormRomanizationGlossTranslation
うちなーぐちuchinaa-guchiOkinawa-mouth‘Okinawan language’ (constructed)
やまとぅぐちyamatu-guchimainland.Japan-mouth‘Japanese language’ (constructed)
しまくとぅばshima-kutubaisland-language‘local language, island speech’ (constructed)
ぬちぐすいnuchi-gusuilife-medicine‘something that nourishes or restores one’ (constructed)
むぬいいmunu-iithing-saying‘speaking, manner of speaking’ (constructed)

In the first two compounds, ‘mouth’ names the speech associated with the place named by the first member. In shima-kutuba, the second member already means ‘word’ or ‘language’. The first member specifies the place or community whose speech is meant. Nuchi-gusui has a conventional figurative reading. Its components make the idea of something restorative intelligible, and the established compound can be used of food or another experience felt to restore or sustain a person.

A compound may preserve the free form of each base or show a different form at the join. Compare the following second members.

Independent formCompound memberEnvironment illustrated
くち kuchi mouth ‘mouth’ (constructed)ぐち -guchi mouth ‘speech’ (constructed)After a place name in uchinaa-guchi and yamatu-guchi
くすい kusui medicine ‘medicine’ (constructed)ぐすい -gusui medicine ‘medicine, restorative’ (constructed)After nuchi ‘life’ in nuchi-gusui
くとぅば kutuba word ‘word, language’ (constructed)くとぅば -kutuba language ‘language’ (constructed)After shima ‘island’ in shima-kutuba

Voicing is visible in the first two rows. The third row shows that a compound boundary alone does not predict it. For an established compound, the form of the second member is part of the lexical evidence. A character or syllable that differs at the join can help identify the formation once the bases are known.

In a noun phrase headed by a compound, particles marking the role of that phrase follow the compound as a whole. A genitive expression, by comparison, has a separately marked dependent inside the phrase. A compound can still correspond in meaning to such a genitive expression. To determine its structure, check whether there is internal phrase marking, whether the members form a conventional unit, and whether either member has a shape associated with the compound. Spacing conventions and a plausible literal translation supply weaker evidence.

Verb compounds

In a verb compound made from two verbal bases, the first can occur in a stem form. The final verb supplies the conjugation of the compound. The first member can indicate an action that leads to the result, or how the final action is carried out. The construction is lexical when the two bases form a conventional predicate and the tense, polarity and other inflectional choices apply to that predicate as a whole.

FormRomanizationGlossTranslation
とぅいあぎゆんtui-agiyuntake.STEM-raise.NPST‘pick up, lift up’ (constructed)

In this example, ‘take’ gives the initial action and ‘raise’ gives the resulting movement. The right-hand verb carries the inflection. A sequence of linked clauses can also describe taking and then raising something. In that construction, the first action has its own clause structure and is expressed in a form licensed for clause linkage. An independently expressed dependent of the first action or separate temporal modification can provide evidence for two clauses. A shared participant and a close connection in meaning are compatible with either structure.

An expression made from a noun and sun also has inflection at the right edge. The class and function of the first member distinguish it from the combination of two verb bases. In yuntaku sun, ‘chat’ names the activity performed. In tui-agiyun, the first member itself supplies a verbal contribution to the event.

Reduplication and expressive formations

Reduplication repeats all or part of a form within a word or a conventional expression. In expressive vocabulary, repeated material can evoke successive occurrences of a sound, movement or sensation. It can also contribute intensity or duration. The contribution depends on the particular word. Repetition gives no general rule for forming noun plurals, verb tense or adjective degree.

FormRomanizationGlossTranslation
あちこーこーachi-kookoohot-EXPR‘piping hot’ (constructed)
ちむどんどんchimu-dondonheart-pounding.EXPR‘heart pounding’, as with excitement (constructed)
にーぶいかーぶいniibui-kaabuisleepiness-EXPR‘drowsy, nodding with sleep’ (constructed)

The repeated koo in achi-kookoo is part of an established expression of intense heat. The property element achi- occurs once. In chimu-dondon, ‘heart’ also occurs once. The repeated don depicts successive beats. The whole expression brings together the body-part noun and the depiction. In niibui-kaabui, the two parts echo one another with a change in sound. Its shape supports an expressive pairing, with variation between the two parts.

A repeated sequence that has no independently established meaning can still be identified as repeated material. Its gloss can describe its expressive contribution without assigning each copy a separate lexical meaning. An independently meaningful base makes a fuller analysis possible when the meaning of the repeated form also follows from that base. For a partially repeated form, the analysis should identify the part that recurs. There is no general instruction to double the first syllable or to repeat the whole base with the same result.

In speech, a speaker can repeat an ordinary word to emphasize it, repair an utterance or call again. Such repetition operates across occurrences of the word in the utterance. Separate intonation and separate opportunities to add particles or dependents can show that the repetition is syntactic or conversational. An established expressive form has a conventional order and interpretation within the expression. Either type can appear as adjacent repetitions in writing.

Loan adaptation and lexicalized formations

Borrowing supplies a lexical form and adaptation integrates it into local usage. Integration can affect the sounds selected, their sequence, the spelling, or the grammatical construction in which the word occurs. In present-day usage, the same speaker can draw on established Central Okinawan vocabulary and on forms familiar from Japanese. A form shared with Japanese may reflect borrowing, common inheritance or a switch of language in that utterance. Its behavior in Central Okinawan phrases and its history provide different kinds of evidence (Chamberlain 1895, Grammar and Dictionary; Shimoji and Pellard 2010, Introduction).

FormRomanizationGlossTranslation
パンpanbread‘bread’ (constructed)
コーヒーkoohiicoffee‘coffee’ (constructed)

These forms illustrate loan vocabulary usable as nouns. Their shape does not entail a derivational suffix. In koohii, the length marks in the kana spelling correspond to long vowels in the romanization. A loan in a noun phrase can occupy the same positions and take the same following particles as other nouns of the same distribution. An action name borrowed as a noun can likewise enter a predicate with sun if its use with that verb is established. In both cases Central Okinawan morphology and syntax apply to the lexical material available to the speaker.

A loan can enter as an entire expression, including parts that have a separate analysis in the donor language. Local derivation is established when the corresponding bases or affixes also participate in the local formation. A plausible donor and a close resemblance in sound can support a borrowing account. The exact route of transmission requires historical evidence. In particular, the current form of a shared word may be compatible with more than one route.

Clipping, blending and lexicalization

Clipping shortens a longer lexical form while keeping its identity as a name for the same thing. It is useful to distinguish a conventional short form from reduced pronunciation in connected speech. A conventional short form can recur independently, including in careful speech, with the reference of the longer form. A reduction restricted to a particular sound environment belongs to connected speech. A compound that uses a combining stem gives evidence for that stem, and evidence of independent use is needed to establish a corresponding clipping.

A blend draws on two source forms and uses only part of one or both, or overlaps material from them. A transparent compound can also be shorter than a phrase with the same general sense. Its members, however, have the shapes expected of its bases or combining stems. To identify a blend, the two proposed source forms must account for both the shape and the meaning of the result. An echo formation such as niibui-kaabui supplies a recurring sound pattern. A blend analysis would additionally require independently identifiable source words whose material contributes to that particular form. Clippings and blends are described item by item in Central Okinawan. There is no general slot in the grammar that specifies a shortening or a point of overlap.

Lexicalization establishes a form with its own conventional use. A lexicalized formation can remain transparent in structure. The bases in nuchi-gusui are recognizable, and its restorative sense is conventional. Other formations may preserve an older shape or a meaning that survives mainly in one expression. The current use of such a word can be described directly. A historical analysis of its parts accounts for its origin, while a productive analysis requires evidence that speakers use the same process in other formations.

How Mova handles derived forms and compounds

Mova has neither an annotated Central Okinawan treebank nor a level-2 parser model for ryu. No treebank baseline has been measured. In tools/mova-parse, the # parse comment reports a missing model. The affected tokens have _ in LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL, and Parse=None in MISC.

Language identification in tools/langseg depends on its script and lexicon coverage. For Central Okinawan, Mova does not perform language-specific tokenization, tagging, lemmatization or parsing. Its frames, pragmatics and world components, prag and world, read English trees. They have no parsed Central Okinawan input to analyze.

Squid reads each grammatical point in the chapter as a claim. Rules and other formalizable claims wait in its queue, and none of this chapter's rules is implemented. The first points relevant to a tokenizer, tagger and UD treebank baseline include the spelling of bound forms, the treatment of compounds and the word class of derived forms. The baseline specified in What every language gets uses en xpos-fill before en train trains the parser and tagger on the UD training split. en parse-eval then measures UAS and LAS on the development split.

Sources for this chapter

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/ja/ryu/16-word-formation/