Kosi Grammar Sketch (Hnaumelnihava)
Robert Jung
7 October 2006

Introduction: Kailsai Phonology and orthography: Halerni serelni Nominal morphology: Tinsaillükelni Pronouns: Teloksaul Quantifiers and numerals: Sminsaul som Verbal morphology: Jatsaillükelni Derivational morphology: Kabarasaillükelni Syntax: Sauldehelni Babel text: Babelkmor Lexicon: Saulhava

Introduction (Kailsai)

I feel obliged to express my sincere gratitude for the three grants given me by the Finnish Ministry of Culture for my endeavours vis-à-vis the Kosi language, as well as towards the following people from the Kosi community: Hilda Kristin Holopainen, Adam Dani Kalliomäki, B. Matu Kääriäinen, Viljam Kristian Haugland, Tomas Petersen, Lisa Valtasaari, and Jan Kristov Nevalainen. These people tirelessly aided me in my quest to learn their language, spending long hours discussing it with me, and allowing me to live with them for many months. I would also like to acknowledge and thank Rory Turnbull for hosting this sketch.

Kosi is spoken by about 250 people in the northern Finnish towns of Utsjokisuu and Dalvadas in the Utsjoki municipality. The language is classified as endangered by the Finnish Ministry of Culture; every year money is allocated for research and documentation purposes. Most speakers of the language are bilingual in Finnish and Kosi, although there are about 30 monolingual native speakers. There are Kosi language classes available in the Utsjoki municipality, which about fifty people attend regularly, including many of high-school age. It seems likely that Kosi will survive for at least a few more generations, given the efforts of local communities and the national government as well as many linguists in preservation and documentation (Carlos Ramírez 1999:64).

No genetic link between Kosi and any other language(s) has been proven to date. I oppose this endeavour because of the wasted resources, which could have been used to continue the study and documentation of the language. Fortunately, the Finnish Ministry of Culture has spent little money on such projects. Kosi has, for the most part, remained intact as a unique entity from Finnish and Sámi; its speakers have strong linguistic pride and actively avoid borrowing from other languages.

The language is a highly agglutinative, dominantly SOV language, with a dozen noun cases; a rich system of aspects, moods, and voices; a large derivational system; and a base-20 number system.

I used the Leipzig Glossing Rules for all interlinear translations in this sketch, with the following modifications and additions: ACT.PTCP active participle, ARB arbitraritive, COM comissive, EXP experiential, HAB habitual, INC inceptive, MOM momental, PASS.PTCP passive participle, PAUS pausative, PRO protractive, RESUL resultative, RESUM resumptive, SUB subessive, SUP superessive. It should be noted that if a period comes between a root and an inflectional suffix, this means that the word is inseparable from that concept, e.g. bat 'friend' versus bet 'friend.ACC'. Unlike bat, which can take other inflections, bet cannot because it is inherently accusative.

Phonology and orthography (Halerni serelni)

Consonants. Kosi has six points and moods of articulation represented in its consonantal inventory: bilabial, labiodental, dental, palatal, velar, and glottal; plosive, nasal, fricative, lateral, rhotic, and glide. (Note that the rhotic is actually alveolar, not dental, unlike all other consonants pronounced in that area. Because there is no phonemic difference between dental and alveolar consonants, they will be lumped together as alveolar sounds in phonemic transcriptions.) Kosi has the following consonants:

Four plosives, two voiceless and two voiced; note that aspiration is not heard even on the phonetic level. /t/, /k/, /b/, and /d/ correspond roughly to the "t", "k", "b", and "d" of "stick", "skin", "able", and "idle". /tj/ and /kj/ are realised as [tS], a voiceless postalveolar fricative, which, without the aspiration, corresponds roughly to the "ch" of "cherry". /dj/ is realised as [dZ], a voiced postalveolar fricative, which corresponds roughly to the "j" of "jar".

Two nasals, both voiced. /m/ and /n/ correspond roughly to the "m" and "n" of "mine" and "nine", respectively. They become syllabic (which correspond roughly to the "-om" and "-on" of "bottom" and "button", respectively) preceding an obstruent and another sonorant word-initially, and following an obstruent word-finally. Syllabic nasals do not assimilate to the place of articulation of a following consonant, however.

Three fricatives, one voiced and two voiceless. /v/, /s/, and /h/ correspond roughly to the "v", "s" and "h" of "vine", "so", and "hit". Some older speakers pronounce the latter consonant as [x], a voiceless velar fricative; the shift to [h] is most likely due to influence from Finnish.

One lateral, which is voiced. /l/ corresponds roughly to the "l" of "light"; note that /l/ is never realised as [l_G], a velarised /l/, which corresponds roughly to the "ll" of "full". /l/ is realised in clusters with obstruents as [K] when the obstruent is voiceless or [K\] when it is voiced, lateral fricatives, which correspond roughly to the "ll" of (Welsh) "llan" and .

One rhotic, which is voiced. /r/ corresponds roughly to the "rr" of (Castillian Spanish) "perro". It turns into a tap (which corresponds roughly to the "r" of (Castillian Spanish) "pero") in consonant clusters and syllable-final position; examples include the /-r-/ in mbatr 'soldier' and the /-r/ in ör 'night'.

One glide, which is voiced. /j/ corresponds roughly to the "y" of "yet".

Vowels. Kosi has three degrees of openness and two degrees of frontness represented in its vocalic inventory: close, close-mid, and open; front and back. (Note that the close-mid front unrounded vowel is actually open-mid in all positions, unlike all other vowels. Because there is no phonemic difference between close- and open-mid vowels, the aforementioned vowel and its less open cousins will be lumped together as close-mid in phonemic transcriptions.) Kosi has the following vowels:

Three close vowels, one unrounded and two rounded. /i/ corresponds roughly to the "i" of (French) "joli". It turns into a central vowel (which corresponds roughly to the "y" of (Polish) "trzy") after /k/ and /h/. /y/, written ü, corresponds roughly to the "u" of (French) "tu". It turns into a central vowel (which corresponds roughly to the "u" of (Norwegian) "kultur") after /k/ and /h/. /u/ corresponds roughly to the "ou" of (French) "tout".

Three close-mid vowels, one unrounded and two rounded. /e/ corresponds roughly to the "e" of (French) "sept". /2/, written ö, corresponds roughly to the "eu" of (French) "feu". /o/ corresponds roughly to the "eau" of (French) "peau".

One open vowel, which is unrounded. /a/ corresponds roughly to the "a" of (French) "chat". It turns into a central vowel (which corresponds roughly to the "-er" of (German) "Werner") in polysyllabic word-final position.

Diphthongs. Kosi has ascention represented in its diphthongial inventory. It has the following diphthongs:

One composed of /a/ and /i/. It corresponds roughly to the "ay" of (Castillian Spanish) "hay".

One composed of /a/ and /u/. It corresponds roughly to the "au" of (Castillian Spanish) "causa".

Minimal pairs. Two sets of minimal pairs proving the phonemic status of all the consonants mentioned above are as follows:

Two sets of minimal pairs proving the phonemic status of all the vowels and diphthongs mentioned above are as follows:

Stress. Primary stress falls on the first syllable of a word and then on any subsequent odd syllables: NAki, nakeNA, etc.; monosyllabic function words are usually unstressed. Elements that the speaker wishes to emphasise have a lower pitch than unstressed ones. Statements have a level intonation, but questions have a rising intonation; the two need to be distinguished somehow as questions are not marked morphologically.

Syllables and phonotactics. Syllables divide before a single consonant (except word-finally): CVCVC > CV.CVC, CCVCVCC > CCV.CVCC; between two vowels: CVVCVC > CV.V.CVC; and after the first consonant of a cluster intervocalically: CVCCVC > CVC.CVC. Kosi syllables can consist of at most two onset consonants, a vowel/diphthong/syllabic consonant nucleus, and two coda consonants; hence, Kosi's syllable structure is roughly as follows: (nonsyllabic consonant) plus (nonsyllabic consonant) plus vowel/diphthong/syllabic consonant plus (nonsyllabic consonant) plus (nonsyllabic consonant), where brackets represent optionality. All single consonants are allowed initially, medially, and finally (in the latter case, excluding /j/). Geminates are allowed only medially. (The rest of this being a bit dull, perhaps you would like to move on?) Possible initial consonant clusters include plosive+nasal, plosive+r, nasal+consonant, liquid+consonant, s+consonant, and consonant+s. Possible final clusters include liquid+consonant, nasal+consonant, consonant+s, and s+consonant. /h/+consonant cannot appear initially or finally unless the consonant is a liquid or nasal. consonant+/h/ and consonant+/j/ are prohibited in final position. Geminate consonants or vowels are allowed as neither onset nor coda, but only in medial position; such clusters are pronounced no differently from their single counterparts. Possible medial consonant clusters include any combination of the above. When two consonant clusters differ in voicing, this conflict is resolved progressively, starting from the first consonant in the cluster onward. Sometimes an impossible cluster arises through compounding or (less often) inflection; to resolve the illegal cluster, the following rules may be applied: /h/+/s/ > /h/+/h/, consonant+/l/ > consonant+/j/, /i/+vowel > /j/+vowel, /e/+vowel > /j/+vowel (with one exception: /e/+/i/ > /e/+/i/), /y 2 u o/+vowel > /y 2 u o/+/v/+vowel, /a/+/e/ > /ai/, /a/+/y 2 u o/ > /au/. When a suffix is added, a root's final vowel disappears.

Nominal morphology (Tinsaillükelni)

Case. Kosi has a number of cases, totaling twelve: seven core cases and five local cases. Nouns ending in -a, -i, and -ü have no distinct dative, instrumental, or associative form respectively, e.g. Kosi 'in Kosi', not *Kosii. This has led to back-formation in some speakers' idialects, where Kos- began to be analysed as a root and Kosi as an inflected form thereof. (Curiously, since Kosi prohibits single-consonant roots, the word aita has been borrowed from Finnish to fill the gap of a missing nominative form for ta 'father'. Despite this, most speakers still avoid aita, preferring the native ta for both nominative and dative forms; this is a case in point of the Kosi's pride in their language and desire to keep it distinct (Ramírez 1999:66); this grammar will use a native word in preference to a borrowed one.)

The core cases are as follows:

The nominative marks the subject of a sentence; it has no particular mark. A sentence illustrating its use is as follows: Mas melsai lanmat ('The boy is learning how to cook'), which breaks down like this: boy-[NOM] learn.how.to cook.

The accusative marks the direct object of a sentence; its usual mark is -t with the allomorph -s after word-final -d and -t. A sentence illustrating its use is as follows: Mas melsai kasolema hibja (meaning 'The child is learning how to cook clowns'), which breaks down like this: boy learn.how.to clown-DAT cook.

The dative marks four distinctions; its mark is -a.

The partitive marks the action as being directed towards only part of the object; its mark is -os; note that root-final -s changes to -r when this ending is added. A sentence illustrating its use is as follows: Las hasaudoloros tab (meaning 'The dog ate part of the homework'), which breaks down like this: dog homework-PART ingest.

The genitive marks a noun as being owned by another (animate) noun, and translates English possessive phrases and benefactives; its usual mark is -u with the allomorph -uh before other case endings.

The associative marks three distinctions; its usual mark is -ü with the allomorph -üh before other case endings.

The instrumental marks five distinctions; its mark is -i.

The fifth use of the dative is more difficult to explain. It indicates that the marked noun(s) is in some kind of locative relationship to the subject; kol 'centre', lauh 'left hand', tani 'right hand', kev 'head', dön 'foot', mkiv 'face', dur 'back', and höv 'side' in the inessive or allative case appear between the subject and the marked noun(s).

The local cases are as follows:

The inessive (-au) marks two distinctions.

The allative (-ain with the allomorph -aim when the root ends in a dental consonant) marks three distinctions.

The ablative (-ö) marks two distinctions.

The superessive (-ab with the allomorph -ad if the root ends in a bilabial consonant) indicates that there is something on top of or above the marked noun. A sentence illustrating its use is as follows: Lan dönab rjoka (meaning 'The girl is lying on the floor'), which breaks down like this: girl floor-SUP lie.down.

The subessive (-isu) indicates that there is something located on the underside of or below the marked noun. A sentence illustrating its use is as follows: Naki nturisu daseh lisivra (meaning 'The child is hiding under the bed because of the thunderstorm'), which breaks down like this: child bed-SUB hide.oneself RESUL-thunderstorm.

Demonstratives. Kosi has three demonstratives. The proximal (-umo) indicates the marked noun is located near the speaker. A sentence illustrating this is as follows: Kasolem janumo maksöm (meaning 'The clown wants to buy this game'), which breaks down like this: clown game.ACC-PROX.DEM COM-buy. The obviative (-at) indicates the marked noun is located near the listener, e.g. Kaivem havat makmot (meaning 'The dictator wants to destroy that book'), which breaks down like this: dictator book-OBV.DEM COM-destroy. The distal (-öt) indicates the marked noun is located far from both the speaker and listener, e.g. Mbatem mbatjöt maksöm (meaning 'The soldier wants to buy yon gun'), which breaks down like this: soldier gun-DIST.DEM COM-buy. Instead of having separate plural forms of the demonstratives, the head noun is pluralised, e.g. Ukrahem olmat makbaksa (meaning 'The thief wants to grab hold of those apples'), which breaks down like this: thief red.apple\PL-OBV.DEM COM-grab.hold.of. Unlike case endings, the demonstratives are actually clitics; they attach to the final constituent of a noun phrase (which they can actually replace). Moreover, there is no nominative/accusative distinction between nouns with demonstrative endings.

Pronouns (Teloksaul)

üs, ket, and rih (from the numbers 'one', 'two', and 'three') are used when dealing with multiple third-person referents. Here they, similar to demonstratives, represent third person proximal, third person obviative, and third person distal.

Possessive pronouns are also absent from the language. The genitive or associative form of a noun is instead used.

The genitive or associative case is used to handle multiple third-person possession.

Quantifiers and numerals (Sminsaul som)

Kosi has a mixed base-10 and base-20 number system. The most basic cardinal numbers are as follows: üs '1', ket '2', rih '3', dser '4', elt '5', söka '6', kuvo '7', hire '8', kri '9', mbak '10', brauk '11', tljo '12', draki '13', oteks '14', laums '15', sürüm '16', radah '17', lumos '18', kibr '19', van '20'. Within a group of 20 (e.g. 21-40), the smaller number comes first, e.g. üsvan '21' (1+20), ketvan '22' (2+20), rihvan '23' (20+3). In forming higher multiples of 20, the number of 20's follows van, e.g. vanket '40' (20x2), vanrih '60' (20x3), vandser '80' (20x4). The next number with its own lexical root is ndal '100', followed by stah '1,000', ils '100,000', setain '1,000,000', and nekonja '1,000,000,000' (i.e., '1,000 million').

Ordinal numbers, which can precede a noun or stand alone, are created with -na, e.g. rihna 'third', dserna 'fourth', eltna 'fifth'. There are two irregular forms: bna 'first' (lit. 'important') and hru 'second' (lit. 'other').

Positive numbers are formed with the augmentative suffix -ös, e.g. ketös '+2'. Negative numbers are formed with the diminutive suffix -üm, e.g. ketüm '-2'. Fractions, which can also precede a noun or stand alone, are created with -um- between the numerator and denominator, e.g. ketumrih '2/3', dserumelt '4/5'. There are three irregular forms: söl '1/2', moh '1/3', nöl '1/4'. Mixed fractions are formed by placing the whole number before the fraction, e.g. üsketumrih '1&2/3', üsdserumelt '1&4/5'.

Verbal morphology (Jatsaillükelni)

Aspect. The aorist (ja-) marks indefinite time and universal truths. A sentence illustrating this is as follows: Harumo jabulat (meaning 'The earth is round'), which breaks down like this: Earth AOR-round.

The habitual (to-) marks a verb as being performed on a regular basis, e.g. Lan srend todosbaksa mkivau el tab (meaning 'The girl juggles cookies before eating them'), which breaks down like this: girl cookie.ACC juggle before them.ACC ingest.

The inceptive (me-) marks an action as beginning, e.g. Man bahadjot mebahad (meaning 'The man is beginning to smoke a cigar'), which breaks down like this: man cigar-ACC INC-smoke.

The momental (usually vi-, with the allomorph ve- before /j/) marks an action as happening for a brief time, e.g. Mas len vilat (meaning 'The boy glanced at the girl'), which breaks down like this: boy girl.ACC MOM-look.at.

The protractive (mik-) marks an action as happening for a long time, e.g. Mas len mikjat (meaning 'The boy stared at the girl'), which breaks down like this: boy girl.ACC PRO-look.at.

The completive (bo-) marks an action as being totally completed, e.g. Man bahadjot bobahad (meaning 'The man is completing smoking a cigar'), which breaks down like this: man cigar-ACC COMPL-smoke.

Mood. The indicative (no particular prefix or suffix) marks an action as being factual. A sentence illustrating this is as follows: Robert jalmös salmort tül (meaning 'Robert likes history'), which breaks down like this: Robert enjoy history-ACC study.

The subjunctive (s- with allomorph d- in roots with initial s-) marks an action as being plausible.

The negative (n-) marks an action as being negated; the placement of the prefix in the verbal complex can change the meaning subtly, because it only negates the morpheme immediately following.

The imperative (nda- singular, nde- plural) marks two distinctions. A sentence employing the first ends in an exclamation point and has an increased volume, the second a period and a level tone.

The comissive/polite imperative (mak-) marks two distinctions. A sentence employing the first ends in a question mark and has a rising tone, the second a period and a level tone.

The permissive (l-) marks an action as allowed to be done, e.g. Kelem vai kedinharat lvitari jadesanaim drohain lidos? (meaning 'May I borrow those watermelons to throw at the hated king?'), which breaks down like this: speaker many watermelon-OBV.DEM PER-borrow AOR-hate-PASS.PTCP-ALL king-ALL RESUL-throw.

Con- and disjunctions. Kosi lacks conjunctions entirely; instead, prefixes on the verb are used. If a serial verb phrase appears, the conjunctive prefix attaches to the last verb.

There is no direct equivalent of English 'and'; juxtaposition usually conveys the proper meaning.

stau- is roughly translatable as 'etc.'; it is similar to 'and' but is only used when there are more possible items that could be listed.

e- has no exact English equivalent; it expresses an exclusive 'or', where only one of the two (or more) choices is possible.

u- translates roughly as English 'and/or', where both (or more) of the choices are possible.

da- translates roughly as English 'but (not)', where the second clause is opposite in negation to the first.

dro- translates roughly as 'although/despite/in spite of'.

vu- translates roughly as English 'in fact', which emphasises or clarifies the previous idea.

Derivational morphology (Kabarasaillükelni)

Nominalisations. Although Kosi lacks simple nominalisations of verbs and adjectives, e.g. kot 'happy/joy', it has a large number of affixes. A noun may denote:

1. a single occurrance of a verb (-is or -üs, depending on the rounding of the final root vowel): bahad 'breathe' > bahadis 'breath', tul 'windy' > tulüs 'breeze'.

2. the overall process of a verb and the name of an ideology (-ira or -üra, depending on the rounding of the final root vowel): bahad 'breathe/be alive' > bahira 'breathing/life', Marks 'Marx' > Marksira 'communism'.

3. the positive result of a verb (-üli): lük 'build' > lüküli 'product', kas 'try' > kasüli 'success'. If the speaker wishes to add a sense of negativity to the result, the adjective smeha 'unfortunate' (often shortened to sme in speech) is added: lük 'build' > smeha lüküli 'worthless piece of trash', kas 'try' > smeha kasüli 'failure'.

4. an actor, (em): tosüt 'prepare food' > tosütem 'cook/chef', kot 'happy' > kotem 'happy person'.

5. a patient (-auk): möt 'injure' > mötauk 'injured person', bat 'loyal' > batauk 'popular person'.

6. an inhabitant of a specific region or area (-r): Sümi 'Finland' > Sümir 'resident of Finland', Helsinki 'Helsinki' > Helsinkir 'resident of Helsinki'.

7. the realisation of a characteristic in the noun (-in or -im for roots ending in -n): maur 'big' > maurin 'big thing', san 'red' > sanim 'red thing'.

8. something concrete associated with a particular action (-ev): tüs 'burn' > tüsev 'fire', nja 'awesome' > njev 'masterpiece'.

9. the most common tool for executing the verb (-in or -jo), e.g. tüs 'burn' > tüsjo 'match', tab 'eat' > tabjo 'utensil'.

10. the usual or expected location of an action or object (-has if the location is a building or -jok if the location is not a building): sadr 'sell' > sadrhas 'store', tül 'study' > tülhas 'school'; dr 'flower' > drjok 'garden', tab 'eat' > tabjok 'kitchen table'.

11. the source plant of some flowers, fruits, and vegetables (-am or -an if the root ends in -m): söl 'grape' > sölam 'grapevine', steni 'cherry' > stenam 'cherry tree', lekm 'wheat' > lekman 'wheat stalk', mlöd 'pea' > mlödam 'pea pod'.

12. a disease afflicting the noun (-kit): kari 'heart' > karkit 'heart cancer', nel 'tongue' > nelkit 'tongue cancer', tisnaki 'bone marrow' > tisnakit 'luchemia'.

Verbalisations. Verbs may denote:

1. a referent is unaware of something that would probably cause him NOT to be doing the action or would affect his attitude towards the situation (-jar). It does not imply that he should have known the missing information; it merely expresses a fact. A sentence illustrating this is as follows: Naki kotjar (meaning 'The child is happy, but doesn't yet know that his favourite toy was thrown away'), which breaks down like this: child happy-ignorantly.

2. a referent feels the marked verb is a chore (-bah), e.g. Naki tabjet tisribah (meaning 'The child is washing the dishes, and thinks it is a chore'), which breaks down like this: child eat-thing\PL-ACC clean-chore.

3. the marked verb is being performed despite expectations to the contrary (-alu), e.g. Naki mara kaumt tisralu (meaning 'The child is washing the dishes, contrary to expectation'), which breaks down like this: child all dish-ACC clean-contrarily.

4. the action, if the evidence is correct, is factual (-ai), e.g. Naki tabjet tisrai (meaning 'The child appears to have washed the dishes'), which breaks down like this: child eat.thing\PL-ACC clean-apparently.

Kosi has no other verbalisers; any derived word was originally a verb. However, some verbs can be used as nouns or adjectives: buk 'read/book', has 'reside/house', jat 'play/game'.

Adjectivisations. Adjectives may denote:

1. an all-purpose active participle (-ak): lat 'see' > latak 'seeing', dat 'not see' > datak 'unseeing'.

2. an all-purpose passive participle (-an): lat 'see' > latan 'seen', dat 'not see' > datan 'unseen'.

3. an abilitative active participle (-ura): lat 'see' > latura 'having sight', dat 'not see' > datura 'blind'.

4. an abilitative passive participle (-eko): lat 'see' > lateko 'visible', dat 'not see' > dateko 'invisible'.

5. an opposite (n-): em 'human-like' > nem 'animal-like', kümr 'wet' > nkümr 'dry'.

6. an opposite which is not quite an antonym (r-): kot 'happy' > rkot 'neither happy nor sad', dsöra 'hot' > rdsöra 'neither hot nor cold'.

Augmentation and diminution. Augmentatives are formed with -ös or -al, diminutives with -üm, -ed, or -si (the latter used for endearment): smin 'amount' > sminös 'a lot', sminüm 'a little'; vilna 'to shine' > vilnös 'to glare', vilnüm 'to shimmer'; ta 'father' > tasi 'dad'.

Sufficiency. Three prefixes exist to indicate whether the amount of a noun or the 'strength' of a verb is sufficient. kot- indicates sufficiency: has 'house' > kothas 'enough houses', mat 'high quality' > kotmat 'adequate'. dri- indicates insufficiency: has 'house' > drihas 'too few houses', mat 'high quality' > drimat 'inadequate'. lan- indicates oversufficiency: has 'house' > lanhas 'too many houses', mat 'high quality' > lanmat 'outstanding'.

Compounding. Kosi is quite free in its allowance of compounding. In compounds of a noun plus noun or a verb plus verb, the more important or significant word comes second. In compounds of a verb and noun, the order is just that. In compounds of a modifier (adjective or participle) and noun or verb, the modifier comes first.

Syntax (Sauldehelni)

Word order. Basic word order is SOV, or Subject+Object+Verb. Indirect objects tend to follow the direct object, and adverbials tend to immediately precede the verb. Despite these general trends, they are not very strict; virtually any argument can be fronted if the speaker wishes to set it apart from the other arguments.

Noun phrases. Prepositional phrases come first, followed by numbers, then adjectives, then participles, then the noun in its inflected glory. Everything except quantifiers and numerals agrees in case and number with the head noun.

Entire noun phrases can be replaced by demonstratives: ut (accusative ol) is used to refer to something the speaker intends to mention very soon, at (accusative el) to refer back to something the speaker just mentioned, and öt (accusative öl) to refer back to something the speaker mentioned some time before. These pronouns can be inflected for other cases, as well, but they have no plural forms.

Verb phrases. Adverbials of place or time come first, followed by derived adverbs (formed with -ai), and then conjugated verbs. There is no agreement between any of those three.

Equational clauses. In an equational clause, neither argument is marked for case, but their ordering has an affect on emphasis.

Intensional clauses. The subordinate clause comes first, followed by the main clause.

Relative clauses. Relative clauses are formed using participles and participle plus noun phrases; relativisation is quite limited, as the relative clause can consist only of these items.

If more specification is needed, the information is presented in the next clause(s).

Causation. The basic causative (ka-) refers to a situation where a person is caused to execute an action; either (a) no resistance is met, or (b) the speaker doesn't want to draw attention to what methods were used to achieve the final result. The agent (the one doing the causing) is in the nominative case, the patient (the one being caused) is in the accusative case, and the second patient (the receiver of the action of the one being caused) is in the dative case.

The persuasive (ki-) refers to a situation where a person is talked into executing the action. The agent (the one doing the persuading) is in the nominative case, the patient (the one being persuaded) is in the genitive case, and the second patient (the receiver of the action of the one being persuaded) is in the dative case. (Note that the persuasive is similar to the last use of the dative case in function.)

The coersive (huk-) refers to a situation where a person is coersed, especially with (threats of) physical force or harm, into executing the action. The agent (the one doing the coersing) is in the nominative case, the patient (the one being coersed) is in the genitive case, and the second patient (the receiver of the action of the one being coersed) is in the dative case.

Voice. The active voice (no particular prefix or suffix) has the subject in the nominative case and focused; the direct object in the accusative case and oblique; and the indirect object in the dative case and oblique.

The passive voice (-ah or -an) has the original direct object in the nominative case and focused, the original subject in the instrumental case and oblique, and the original indirect object in the dative case and oblique; or the original indirect object in the nominative case and focused, the original subject in the instrumental case and oblique, and the original direct object in the accusative case and oblique.

The reflexive-reciprocal (-eh) causes no quirky case behaviour; the singular is always reflexive, whereas the plural is usually reciprocal except for verbs of personal care such as tisreh ‘wash oneself/each other’. If the lack of object causes ambiguity, the speaker must specify further, as there is no morphological way of differentiating between the two.

Similes. Similes and the equative form of comparison are formed with the suffix -ai.

Questions. Kosi has two types of questions: those that request affirmation or negation (yes/no questions) and those that request specific information (non-yes/no questions). Having no words for 'yes' or 'no', the verb tsin 'do' is used as a placeholder (for emphasis, the original verb may be used instead, but tsin is more frequent to prevent repetitiveness). Requests for specific information are formed with the interrogative pronoun, kim; it functions as a placeholder for the desired information, and is not fronted as it is in English.

Existance and possession. The verb neb 'exist' is used to translate 'there is/are' and 'have'.

Comparison. Kosi has no way of marking comparison morphologically; periphrasis is used instead.

Comparison to a previous time is also possible, accomplished with the inceptive aspect (me-).

Conclusion (Hrasaul)

If you have any comments or criticisms of this sketch, feel free to contact me at this e-mail address:
robertmjung AT hotmail DOT com