Kuyʔūn/Morphology

Verbs

The verbal system of Kuyʔūn was mostly based on the categories and affixes inherited from Adāta: Mood, polarity, and emphasis (the latter derived from the Adāta preverbal particle) were marked with prefixes, while aspect and number were signified through a single fusional suffix. The morphological distinction between active and passive voice, however, had been leveled through sound changes, and so new voice suffixes were introduced through verb compounding (a development shared with Mavakhalan).

A unique feature of the Kuyʔūn verb, setting it apart from all other Dāiadak languages, was the advent of a classifier system which can only be explained with deep-rooting influence from native Habeo speakers learning Adāta as a second language.

The morphological template for Kuyʔūn verbs was as follows:

(emphasis)

polarity/mood

(classifier)

verb stem

voice

aspect/number

Most Kuyʔūn verbs had two different stem forms: the so-called primary stem, often ending in a consonant, which was used for the habitual aspect, and the secondary stem, marked by an additional stem vowel and sometimes also featuring consonant mutations, which formed the basis for the perfective and progressive aspects.

The citation form is the primary stem, which is usually identical to the habitual singular indicative active with no classifier. Secondary stems will be given in the lexicon for all verbs.

Mood, polarity, and emphasis

Kuyʔūn inherited most of its mood prefixes from Adāta; however, the benefactive and futilitive moods were lost, and a new subjunctive mood was created from the quotative particle rē.

The irregular negative forms of consonant-initial modal prefixes were analogically prefixed with m- at some point. As a result, the reflex of the affirmative particle ro, which had turned into an obligatory prefix for all positive verb forms, became morphologically redundant and was reanalysed as an optional emphasis marker. Only the optative mood retained traces of this marker in the regular modal morphology.

After this, the Mešmo and Aylatu dialects began using the emphasis marker as a fully independent affix, applying it even to negative forms. (It remains a compulsory affirmative-only marker in the dialect of Cexotúri.)

Most modal prefixes had different variants depending on whether they were followed by a vowel or a consonant. Because of this, and because the negative marker m- sometimes replaced material from the corresponding affirmative marker, the combined affixes will be given in the table below.

	affirmative					negative
	ordinary		emphatic			ordinary		emphatic
	_C	_V	_C	_V		_C	_V	_C	_V
indicative	Ø-		lo-	l-		ma-	m-	loma-	lom-
imperative	yay-		lay-			may-		lomay-
optative	lo-	lok-	lolo-	lolok-		mo-	mok-	lomo-	lomok-
obligative	so-	s-	los-			mas-		lomas-
conditional	po-	pal-	lopo-	lopal-		mapo-	maʔl-	lommo-	lommal-
subjunctive	lī-	liy-	lolī-	loliy-		mī-	miy-	lomī-	lomiy-

A few of these prefixes had predictable variants in certain situations:

• The indicative negative prefixes ma- and loma- changed their final vowel into e when followed by one of /ʦ ʧ ʃ pj tj kj/ and some instances of /n l/, notably before the flat classifier.
• The final /s/ of the obligative prefixes became ʔ before /m n/, š before /ʧ ʃ/, and coalesced with /j l/ into š ł.
• The imperative prefixes changed to iʔ-, liʔ-, miʔ-, lomiʔ- before /i iː/.
• Similarly, the final /j/ of the subjunctive prefixes became ʔ before /i iː/, causing partial homophony with the imperative.
• The long /iː/ of the preconsonantal subjunctive prefixes became ē before /q/. Also, these vowels were shortened if the following syllable received primary stress.

Classifiers

At any point in time between c. 400 and 1200 YP, native Habeo speakers made up between one and two thirds of the population in the Xōron towns. The economical and political elite, however, consisted of ethnic Dāiadak - and so the natives were required to learn Adāta if they wanted to participate in urban business. Since the Habeo languages were highly synthetic, incorporating various types of adjectivials into nouns and adverbials or even other full verbs into the main verb of a clause, it was only natural for them to emulate the Habeo classifier system by forming noun-verb compounds with suitable Adāta lexemes. Over the course of time, this habit was adopted by more and more people of Dāiadak origin, and the most common nouns used for compounding were solidified into a more or less fixed set of grammaticalised classifiers.

Verbal classifiers in Kuyʔūn were used in a way similar to incorporation. They could reference patients, themes, instruments, or less often, the time or location of the action. As all of these could either be left out or remain in place as an overt argument of the verb, classifiers provided a flexible way to reduce valency as necessary. Transitive verbs without a classifier were assumed to reference a discourse-proximate human, usually in the role of direct object; if all non-subject participants were non-human, a classifier was mandatory. An exception was formed by verbs whose object was a subclause, especially an abstract statement; such verbs did not use classifiers for the most part. A single verb could carry only one classifier at a time.

Depending on whether they were followed by a vowel or a consonant, many classifiers had different allomorphs. The table below lists the standard set of classifiers:

	gloss	_C	_V
a slender rigid object	stick	-was-		< basa "staff"
a slender flexible object	string	-ye-	-yey-	< dei "finger"
a bulky object	round	-ʔemas-	-ʔenn-	< hemaza "stack"
a flat object	flat	-ni-	-ny-	< neia "leaf, sheet"
a container	bowl	-ałi-	-aʔl-	< rather "jar"
a fluid	liq	-to-	-t-	< tō "lake, body of water"
solid food	food	-may-	-nk-	< mik "bread, meal"
a tool or weapon	tool	-yeš-		< iāsi "tool"
a time or location	time/loc	-ʔi-	-ʔiy-	< heia "space, room, area"
an animal	animal	-iš-	-ič-	< eika "animal"
a human opponent	foe	-yap-		< dāpu "warrior"
a human friend or supporter	friend	-yo-	-yow-	< iēbu "believer, follower"
a human (neutral)	-	-Ø-

Some predictable interaction with neighbouring morphemes based on the choice of classifier:

• Stem-initial /t k ʔ s/ became c č š š when preceded by a string, flat, tool, or time/loc classifier.
• Stem-initial /i iː/ became e ē when preceded by a string, flat, or time/loc classifier.
• Unstressed stem-initial /ɑ ɛ i ɑj/ became o o oy oy (preserving underlying length) when preceded by a liq or foe classifier, with /oj/ further changing to ē if the stem-initial syllable was closed.
• The prevocalic variant of the food classifier changed to mik- when not preceded by another prefix, i.e. in the affirmative indicative.
• The tool and friend classifiers selected the prevocalic allomorph of preceding mood/polarity prefixes instead of the expected preconsonantal one, inserting an epenthetic i if the prefix contained no vowel of its own. The final /k/ of the optative prefixes combined with the initial onglide of these classifiers to become č.
• Classifier-final /i/ became e before stems beginning with /q/.
• The final /p/ of the foe classifier became ʔ when followed by one of /m n l/.
• The final /s/ of the stick and round classifiers became ʔ when followed by one of /m n l/, and š when followed by one of /ʧ ʃ/.
• The final /ʃ/ of the animal classifier became s before /ʦ s/. The final /ʃ/ of the tool classifier became s as well when followed by [ʦ] from underlying /t ʦ/, but not when followed by /s/.
• The initial /i/ of the animal classifier became e when preceded by a subjunctive prefix.

Voice

Through sound change, most forms of the Adāta passive fell together with the corresponding active forms. Instead, periphrastic constructions based on a compound with the verb eula- "to take" were used to convey passive meanings (as in Mavakhalan). A reflexive voice was created along the same model from xāia- "to share", a causative voice from thaza- "to make, to do", and a passive-permissive voice from saphi- "to give".

Voice suffixes were normally added to the secondary stem of the verb. The voice markers themselves in turn behaved like normal verb stems, taking mostly regular aspect/number suffixes.

active	-Ø-
passive	-ōla-
reflexive	-ʔaya-
causative	-łala-
permissive	-spi-

Only a few irregularities occurred with regard to voice marking:

• The habitual singular forms of the passive, causative, and permissive voices were -ō, -łau, and -spo respectively.
• The passive and causative voices formed the habitual plural with the alternate suffix -ti, giving the combined markers -ōti and -łasti.
• The long /oː/ of the passive voice caused deletion of preceding short unstressed /ɛ ɑ o/. Preceding short /i/ changed into y after single consonants, combining with /t k ʔ s/ to become c č š š. In all other cases where the passive marker was added to a vowel-final stem, an epenthetic consonant was inserted - y after /i iː ɛ ɛː/, and ʔ after /ɑ ɑː o oː/.
• The initial /ʔ/ of the reflexive voice was dropped after a consonant.
• The initial /ɬ/ of the causative voice hardened to t with dropping of the stem vowel if the last consonant before that was intervocalic /s ʃ m n l j w/, with assimilation of /m/ to /n/, and mutation of /l/ into either /w/ or /s/ depending on etymology; /s/ being the more common outcome.

Aspect and number

The fusional verbal endings which encoded both aspect and number were inherited from Adāta almost unchanged. Only some minor adjustments occurred, such as an early regularisation of the habitual singular which came to use the full stem instead of the shortened special form, and some generalisation of stem vowels and/or suffixes based on the passive inflections where active forms had become irregular, preventing these forms to merge with the more regular passives.

The basic aspect/number suffixes were as follows:

	singular	plural
habitual	-Ø	-łi
perfective	-n	-o
progressive	-š	-a

Morphological variation in aspect/number marking:

• The habitual plural ending had the alternate form -ti, which was used when the last consonant of the secondary stem was an intervocalic instance of /s ʃ m n l/ (even though the primary stem was used for the hab.pl!). /-mti/ would assimilate to -nti, and /-lti -wti/ would usually (but not always) mutate to -sti.
• If the primary stem ended in /ɬ/ followed by an unstressed vowel, the vowel would be dropped to yield geminate łł.
• The perfective plural ending had the alternate form -me, which was used when the last consonant of the stem was a nasal. The stem vowel was dropped in this case, and the preceding nasal assimilated to form a geminate /mm/.
• If the stem vowel was one of /ɑ o/, it combined with the perfective plural ending to become -au. Similarly, the perfective plural ending coalesced with unstressed stem-final /ɛ/ into -eu, and with unstressed stem-final /i/ into -yo (-iyo when preceded by a consonant cluster, and -o when preceded by one of /ʃ ʧ/).
• The progressive singular ending changed the stem vowels /ɑ ɛ/ into i, with the exceptions that stressed short /ɑ/ became e instead, and that any sequence of /ji/ would dissimilate to /jɛ/.
• The progressive plural ending coalesced with the stem vowel /ɛ/ to become -i (-e after /j/) and with the stem vowel /i/ to become -ay. The stem vowel /ɑ/ was deleted before this suffix. After the stem vowel /o/, an epenthetic w was inserted. If the secondary stem ended in stressed /ɛ/ or /i/ (regardless of length), the aforementioned vowel changes did not apply, and an epenthetic y was inserted before the perfective and progressive plural endings instead.

Non-finite forms

Kuyʔūn had an adjectivial participle that described the subject of a verb as being involved in the action. It was formed from the primary verb stem by adding -iyen after a consonant and -yen after a vowel. Inflection for voice was possible, the forms were -ōliyen (pass), -ʔayen (refl), -łaliyen (caus), and -spiyen (perm) respectively. The participle could optionally be preceded by a classifier and/or by the negative marker ma-/m-; however, inflections for emphasis, mood, aspect, and number were not used.

In addition to this, Kuyʔūn could also refer to the action as such by means of a verbal noun, which was formally identical to the participle, except that it could also inflect for emphasis and mood. The verbal noun could then take all standard nominal inflections, with possessive affixes referring to the subject. It was fairly common in speech because all verbs in relative clauses or adverbial phrases were inflected as a verbal noun.

The copula

The inflection of the Kuyʔūn copula was highly irregular, and somewhat defective as it did not inflect for voice or emphasis and could not take classifiers either. All extant affirmative forms are listed in the table below; negatives could be formed by replacing initial /l/ with m in the optative and subjunctive moods, prefixing mi- in the imperative, and prefixing ma- before consonants and m- before vowels in the other paradigms. The only exception was the indicative progressive singular, whose negative form prefixed me-.

	singular				plural				verbal noun
	habitual	perfective	progressive		habitual	perfective	progressive
indicative	ałe	a	ši		ay	pe	ałi		ecen
imperative	yay	ya	yeš		yay	yau	yeh		yayen
optative	lōq	loka	lōš		lōłi	lawe	lōh		lōčen
obligative	so	son	soš		sołi	sowe	soh		seyen
conditional	pō	pōn	pōš		pōłi	pawe	pōh		payen
subjunctive	lī	līn	līš		līłi	liwe	līh		līyen

Nouns

Kuyʔūn nouns inflected for four categories: case, number, definiteness, and possession. Case and possession were marked with agglutinative affixes (which were beginning to show some degree of fusional behaviour), while number and definiteness were conveyed by means of a single fusional affix derived from an article system formed during the Late Vulgar Adāta period.

The morphological template of Kuyʔūn nouns looked like this:

case

number/definiteness

noun stem

(possession)

Proper names did not inflect for number or definiteness - they were always considered definite -, but could take case and possession markers like other nouns. Case prefixes applying to proper names are orthographically separated by a hyphen, with the name itself capitalised: yo-Mešmo "to Mešmo".

Number and definiteness

Like its three geographically closest relatives Mavakhalan, Kozado, and Sawîyaran, Kuyʔūn developed a system of obligatory articles, which inflected for number and definiteness of their referent. In the XVA period these particles began to cliticise to their head noun and eventually became grammaticalised as fusional prefixes. The definite article was derived from the Adāta demonstrative zē (sg) / zēk (pl); the indefinite plural originated in a reanalysis of the collective prefix ā-. The indefinite singular was unmarked.

There were a number of mass nouns which had no singular forms, and collectives which were morphologically singular even if they had plural meaning. Apart from the fact that a part of their paradigm was missing, the inflection of such nouns was generally regular.

In addition to this, nouns preceded by a demonstrative did not take definite inflections.

The basic forms of the number/definiteness prefixes in the dialect of Mešmo were:

	singular	plural
indefinite	Ø-	a-
definite	li-	lik-

Depending on the shape of the word stem, some of these prefixes may select different allomorphs and/or cause a mutation of stem-initial consonants. For the most part such alternations were predictable; however, certain historical mergers rendered some of the conditions opaque. The selection of allomorphs can be summarised as follows:

• The indefinite plural prefix changed to i- before stem-initial /ʦ ʧ ʃ pj tj kj/ and some instances of initial /n l/. It coalesced with stem-initial short /o/ into au- and with stem-initial /ɛ i iː/ into ay-, with stress shifting to the first part of the diphthong if necessary. Before all other stem-initial vowels, the allomorph aʔ- with an epenthetic glottal stop was used.
• The definite singular prefix deleted stem-initial short /ɛ/, reduced to l- before /i iː/, and added an epenthetic glide before all other vowels, giving liy-. Before /q/, the prefix as a whole became le-.
• The prefix-terminal /k/ of the definite plural marker became ʔ before /m n l/, č before /j i iː/ (with /j/ being dropped) and before stressed preconsonantal /ɑj/, and turned into gemination of stem-initial plosives. Before stem-initial /ʔ/, the whole prefix changed into laq-, causing deletion of the glottal stop and triggering some rather unpredictable changes in immediately following front vowels.
• Both definite prefixes could change their initial /l/ into a different consonant when case prefixes were added; see below for details.

Case

Five of the most frequently used prepositions of Adāta had turned into clitics early on, fusing to their head nouns and eventually ending up as case prefixes. It is thought that the development of mandatory topic-fronting helped in this process, as the proclitic adpositionals could not be left stranded in their normal syntactic position later in the sentence.

Case prefixes preceded any number/definiteness markers, and interacted with these to create some allomorphic variation. The basic case markers for Mešmo Kuyʔūn were:

direct	Ø-	i.e. subject and direct object
allative-dative	ayl-	i.e. "to, for"; < īla
locative-temporal	om-	i.e. "at, on, in"; < ob
ablative-partitive	a-	i.e. "from, of"; < ha and/or ax (etymology unclear)
instrumental	at-	i.e. "with, using"; < ate
relational-causal	lo-	i.e. "concerning, because of"; < zō

Occurring morphological complications:

• The allative-dative prefix ayl- had the allomorphs yo- before plosives and nasals in the indefinite singular, and el- before almost all other consonants, including the /l/ of the definite prefixes. Stem-initial /j/ in the indef.sg was, however, deleted after this prefix, which retained the regular form ayl- in that case.
• The final /m/ of the locative-temporal prefix assimilated in POA to following consonants other than /j/, causing /ʔ w l ɬ/ to mutate into /m m n t/ (including the /l/ of the definite prefixes).
• The ablative-partitive prefix a- coalesced with the definite prefixes into assi-/assik-. With the i-allomorph of the indefinite plural prefix it combined to form ay-; with the regular a-allomorph the ablative-partitive prefix was contracted to a single long ā-, which was shortened if the following syllable carried primary stress. In the indefinite singular, an epenthetic ʔ was inserted before stem-initial vowels.
• The final /t/ of the instrumental prefix at- shifted to ʔ before /m n l/, turned into gemination of following plosives, and coalesced with stem-initial /ʔ/ to form ł. Before vowels, the prefix as a whole became ec-.
• The relational-causal prefix lo- added an epenthetic ʔ before vowels, and caused the definite plural prefix to mutate into -sk-/-šč- before stems beginning with a stressed vowel.

Possession

Almost all Dāiadak languages show some kind of possessor marking on possessed nominals, and Kuyʔūn was no exception. As in all other known direct daughters of Adāta, possession was indicated by suffixes derived from cliticised possessive pronouns. Unlike most of its sister languages, however, Kuyʔūn used a regularised pronoun set in which the classical plural forms were replaced by appending -k to the singular pronouns in analogy with the direct case. Also, separate forms for inanimate 3rd person referents were innovated from the demonstrative xaxe "that thing", and a 1st plural inclusive was created from the phrase i on do "I and you". (Compare the section on pronouns for the reflexes of free-standing non-possessive forms of this paradigm.)

The following table lists the basic possessive affixes for the Mešmo dialect:

	singular	plural
1st exclusive	-i	-iš
1st inclusive		-yonno
2nd	-yo	-yoq
3rd animate	-k	-kaq
3rd inanimate	-ʔe	-ʔek

Notable irregularities were as follows:

• The 1st exclusive endings coalesced with stem-final vowels to form a diphthong; in the plural the result was further contracted into a long monophthong, giving the suffixes -īš with stem-final /i ɛ/, and -ēš with stem-final /ɑ o/.
• All non-3rd person forms could cause palatalisation on stem-final consonants, turning /t k q s h/ into c č k š š with dropping of the suffix-initial onglide. /t k/ were usually immune to palatalisation in the 1st exclusive, however.
• The 3rd animate singular suffix had the alternate forms -ak with all stress-shifting nouns, and -ka after consonants other than a nasal.
• After stem-final nasals, the 3rd animate suffixes became -na and -naq respectively, with assimilation of the preceding nasal into a geminate /nn/.
• The suffix-initial /ʔ/ of both 3rd inanimate markers metathesised with stem-final nasals, deleted stem-final /p k q/, and disappeared with gemination of stem-final /s ʃ/. With stem-final /t/ it coalesced into ł, and with most but not all stem-final /w/ it coalesced to form ʔl. Adding 3rd inanimate suffixes to stem-final /h/ normally gave ss, but could also result in ʔl or ʔ depending on etymology. In addition, some vowel-final stems selected the ʔl allomorph instead of the expected simple glottal stop.

Some nouns had irregular stems for use in conjunction with possessive affixes; these are given in the lexicon.

Pronouns and determiners

Personal pronouns

Pronouns in Kuyʔūn inflected for case and possession using the same markers as nouns, but did not decline for definiteness. However, the case-marking system for pronouns was complicated by the fact that the Adāta distinction of direct and oblique pronouns was retained in the singular, not only as a grammatical category but also in the stems used to build the cases shared with ordinary nominals.

The Xoronic dialect of Adāta retained the original 2nd person plural pronoun dok (just like the dialects of Thāras and Lasomo, which eventually developed into Ayāsthi and Æðadĕ). As mentioned in the section on possession markers above, the pronoun system was also extended by the innovation of new 3rd person inanimate and 1st person plural inclusive forms.

	1sg	2sg	3sg.anim	3sg.inan		1pl.excl	1pl.incl	2pl	3pl.anim	3pl.inan
direct	i	to	a	ʔah		ik	yonno	toq	aq	ʔaʔek
oblique	in	ton	an	ʔaʔen
allative-dative	aylin	yoton	aylan	ʔeʔlen		aylik	aylonno	yotoq	aylaq	ʔeʔlek
locative-temporal	omin	onnon	oman	onnaʔen		omik	omyonno	omyoq	omaq	onnaʔek
ablative-partitive	ay	ayo	ā	aʔah		īš	ayonno	ayoq	āq	áʔāk
instrumental	eci	eco	eca	ałah		ecik	econno	ecoq	ecaq	ałaʔek
relational-causal	loy	loyo	lā	loʔah		lēš	loyonno	loyoq	lāq	láʔāk

Correlatives

	query	anaphor	this	that	some	no	every	each
demonstrative	yo		ʔa	ši	nam	ma	eu	ō
person	išo	leyo	ʔayo	šiyo	nanno	meyo	ešo	ōyo
thing	ʔeu	leye	ʔaʔe	šeye	naʔe	maʔe	eʔe	ōʔe
place	mau	lilo	ʔalo	šilo	nalo	malo	elo	ōlo
time	sō	liso	ʔaso	šiso	naʔo	maso	eso	ōso
way	ip		ʔap	šip	nem		ešip
reason	ayla		ʔało	šiło	nin	mało

• The initial consonant correspondences of the deictics were regularised; this included *seye > šeye ("that thing"), and the innovation of new proximal forms ʔaso, ʔap, ʔało (corresponding to English "now", "thus" and "therefore").
• A separate "person" series was formed from iēbu "believer; i.e. person of Anaitist faith" and/or idō "nobleman" (etymology unclear).
• A new series of distributive pro-forms was created from or "each"; the regular pattern suggests an analogy-based origin.
• The "anaphor" series functioned in a similar manner to deictics; however, anaphoric pronouns could only reference a previously established sentence topic. With the "person" and "thing" anaphors, the meaning was often weakened, sometimes comparable to the semantic vagueness of a mere definiteness affix. On the other hand, the "place" and "time" anaphors were always co-circumstantial in meaning: "at the same place/time".

Some of the above correlatives could be pluralised; a few of the formations were irregular. The existing plural forms are given in the following table:

	query	anaphor	this	that
demonstrative	isk		ʔaq	šik
person	išoq	leyoq	ʔayoq	šiyoq
thing	ʔesk	līk	ʔaʔek	šīk

The demonstratives yo, ʔa, and ši agreed in number with the nouns they determined. The singular-only demonstratives nam, ma, eu, and ō were only used with singular nouns; where they apparently referred to nouns in the plural, the particles were in fact not demonstratives, but quantifiers.

Adjectives

Adjectives in Kuyʔūn could either follow their head noun as a separate word, or be incorporated into the noun via head-final compounding. The latter was more common and seen as the unmarked construction for single adjectives; however, only one adjective at a time could be incorporated, while all additional adjectives had to be placed after their head noun. Also, incorporated adjectives could not be cast in the comparative, and there were some lexical restrictions as to which adjectives preferred the incorporated or independent constructions respectively, which meddled with the usual convention to incorporate the semantically most salient adjective only.

Incorporation

Incorporated adjectives were placed right before the noun stem, being subject to the usual morphophonological interaction with any preceding prefixes. However, the boundary between the adjective and the noun stem underwent only minimal phonetical changes:

• Adjective-final unstressed vowels were dropped before noun stems beginning with a vowel unless they could either form a diphthong with noun-initial short /i/ or /o/, or the noun began with an accented short vowel of the same quality, in which case the pair of vowels coalesced into a single long vowel.
• After adjective-final diphthongs or accented vowels, an epenthetic ʔ was inserted if the noun stem began with a vowel.
• Adjective-final /p t k q/ reduced to ʔ before noun-initial /m n l/.
• Adjective-final /ʦ ʧ/ became s š before any noun-initial consonant.
• Adjective-final /m n/ assimilated in POA to noun-initial consonants other than /ʔ j w/.
• Adjective-final /h/ became ʔ before a vowel, and disappeared with lengthening of preceding short vowels when followed by a consonant. This lengthening was blocked if the first vowel of the noun was stressed or long.
• Noun-initial /ʔ/ metathesised with immediately preceding adjective-final /m n l/.

Independent adjectives

Independent adjectives followed their head noun, usually without any intervening morphemes save for possession markers. However, number agreement permitted some shuffling of word order, and so adjectives could occasionally be topic-fronted without their referent.

Plural marking on independent adjectives was accomplished with the suffix -k, which became -q after the back vowels /ɑ o/, coalesced with diphthongs in /-w/ into -ōq, and became -aq after all voiceless consonants except /h/, which was instead deleted with compensatory lengthening on preceding short vowels.

A comparative could be formed from the regular adjective by suffixing -neu (singular) or -nēk (plural). Before this suffix, adjective-final /p t k q h/ reduced to ʔ, /ʦ ʧ/ became s š, and /m/ assimilated to n.

A morphological superlative did not exist; however, a superlative construction could be formed by nominalising the -neu-suffixed adjective and modifying this by a postposed noun in the ablative-partitive case.