Proto-Koyanic language

Proto-Koyanic (PKn) is the reconstructed common ancestor of the Koyanic languages, a language family found throughout the Koyan region of western Fountainhead. Proto-Koyanic was likely spoken as a singular language in Kóty c. 4,000 - 3,000 BGS before breaking up into its major branches via major migrations through the third and second millennia BGS.

Consonants
The reconstructed consonant inventory of Proto-Koyanic is presented below.

Sibilants
The sibilants *z *s pattern very similarly to the plain-aspirated plosive pairs, leading to some suggesting affricate qualities [ts tsʰ] or even palatalised plosives [tʲ tʲʰ]. Nonetheless, the reflexes of these consonants in descendent languages strongly support their reconstruction as a voiced-voiceless pair of fricatives [z s], in particular the frequent rhotacisation of *z to [r] and debuccalisation of *s to [h].

It is quite possible that these consonants developed from older plosives which shifted to fricative qualities, a change which notably occurred again in Proto-Gäj which lost original PKn *z *s, then shifted *c *cʰ to *z *s which continue to pattern like plosives.

Palatals
The palatal plosives *c and *cʰ are reflected by Proto-Duodoseic *k, indicating that Duodoseic may have branched at an early stage of the language which instead had palatalised velar stops *kʲ *kʲʰ, while the rest of the language family strongly supports the reconstruction of true palatals. By convention, these consonants are transcribed *c *cʰ even in considering Proto-Koyanic as the common ancestor of the entire family including Duodoseic.

Laryngeals
* ʡ and *h likely had the same place of articulation and formed a plain-aspirated pair like the plosives. Whether these were pharyngeal [ʡ ħ] or glottal [ʔ h] is not known, but the Duodoseic reflection of *ʡ as a uvular *q could indicate it was pharyngeal, since ʔ > q might be an impossible sound change.

In addition to the inventory above, Proto-Koyanic is reconstructed with a set of three laryngeal consonant-like sounds, *x₁ *x₂ *x₃, which almost always occur next to a schwa *ə *ə̄ as part of a syllable nucleus. These laryngeals colour the adjacent schwa in characteristic ways, though the exact resulting vowel quality varies by daughter language. *x₁ gives a close front colouring [ɨ~i~e], *x₂ an open front colouring [ə~e~æ~a~a], and *x₃ a back rounded colouring [ɑ~ɔ~o~u]. The laryngeals are retained in certain phonological environments in some daughter languages with the following reflexes: *x₁ by [x~ɣ~h~ʔ], *x₂ by [χ~ʁ~h~ʔ] and *x₃ by [x~ɣ~ʁ~h~ɦ]. Since there is such a large overlap in their qualities, not much can be said about the original qualities of each laryngeal, only that they were likely fricatives articulated further back than the velum and *x₃ probably had some labialisation.

Voiceless sonorants
The voiceless sonorants are allophones of their voiced counterparts when in a consonant cluster containing *h or following an aspirated plosive or *s. They are not thought to have been phonemic in Proto-Koyanic, but became consonants in their own right in the Qʼachkav and Hnäzb language groups. The distinctive qualities of the voiceless sonorants are nonetheless reconstructible, and are commonly transcribed ⟨m̥ n̥ ɲ̊ ŋ̊ r̥ l̥ ẙ w̥⟩ to make this clear. When clustering with *h, they are often transcribed as single consonants, e.g *hm̥ə́x₃l̥h "blood" may be written *m̥ə́x₃l̥.

Vowels
It has been suggested that the vowels *u *ū and *a *ā had qualities [ʊ ʊː] and [ɑ ɑː] respectively, though this is difficult to reconstruct with such certainty.

Sonority Hierarchy
The syllable structure of Proto-Koyanic is based on the sonority hierarchy X > G > L > N > O. Glides G and liquids L also form a natural group R in the syllable structure. The combination of groups L, N and O (i.e. all consonants except laryngeals and semi-vowels) is given the cover symbol C.
 * 1) Laryngeal X = {*x₁ *x₂ *x₃}
 * 2) Glide G = {*y *w}
 * 3) Liquid L = {*r *l}
 * 4) Nasal N = {*m *n *ɲ *ŋ}
 * 5) Obstruent O = {*p *t *z *c *k *q *ʡ *pʰ *tʰ *s *cʰ *kʰ *qʰ *h}

Syllable Structure
The maximal syllable structure is *(C)(N)(R)(X)V(X)(R)(N)(C), with a few additional constraints:


 * A laryngeal X may only be present if the vowel V is a schwa *ə *ə̄, and only one laryngeal may be present. There exists a single exception to this rule, the root *√x₁us "bird", a common root which appears in multiple stems and of which evidence abounds that this is the most appropriate reconstruction.
 * The boundary consonants C may be an obstruent or a nasal. In the case of a nasal-nasal cluster, exactly one of the nasals must be *m. For example, *ŋm- and *mn- are allowed (and attested) onsets, while *ɲn-, *nŋ- and *mm- are not.
 * Certain combinations of obstruent and sonorant are forbidden in the onset, as shown in the section below.

There is also a type of minor syllable which has no vowel and only a single laryngeal *X as its nucleus. These appear mostly in roots and suffixes and are known as light syllables, in contrast with heavy syllables shown above, in reference to their differing prosodic weights. A light syllable can only occur following a heavy syllable, and can never carry stress. If a light syllable ends up with no preceding heavy syllable, it is promoted to a heavy syllable by insertion of *⟨ə⟩ after its nuclear X. e.g. *√x₁t “the wind, life” surfaces as *x₁ət- in *x₁ə́t.cəm (H́.H) “the wind god” and *x₁ə́t.mnx₃ə (H́.H) “to blow”, but remains *-x₁t in *x₂əm̥.px₃ə̄́.kəx₃.sx₁t (H.H́.H.L) “to kill”. Note that this process is not related to stress whatsoever, but it does allow an underlying light syllable to take the stress if it surfaces as a heavy syllable.

Onset clusters
As well as the constraints determined by the sonority hierarchy, there are a few other restrictions on which consonant clusters were possible in the syllable onset:


 * Plosive-nasal onset clusters were allowed only when heterorganic, that is the two consonants have different places of articulation. Thus, clusters like **kŋ- and **tʰn- were forbidden while *cm- and *ʡɲ- are allowed.
 * Certain clusters with one palatal consonant an another with a similar places of articulation are also thought to have been forbidden, i.e. **k(ʰ)ɲ-, **t(ʰ)ɲ-, **sɲ-, **zɲ- **c(ʰ)n and **c(ʰ)ŋ.
 * Nasal-nasal clusters were even more restricted, being allowed only when exactly one of the two nasals is *m, e.g. *mn-, *ɲm- and *ŋm- are all attested.
 * Medial *-r- and *-y- only appear following labial, uvular and laryngeal consonants

The table below shows an example for each attested biconsonantal cluster within the onset of a single morpheme in Proto-Koyanic. Greyed out cells indicate clusters thought to be phonotactically forbidden, while the other empty cells are simply not attested. Fortition was a property applied to the entire onset cluster and is not phonotactically relevant, so the fortis-lenis obstruent pairs are considered together.

Stress
Stress usually falls on the penultimate heavy syllable, but it’s possible to fall on other heavy syllables which have specific morphologically marked stress, e.g. in certain suffixes.

Roots
Roots are a particular morphological unit which can act as a word stem with a class just like any other stem, but are most typically used to derive new stems. The canonical Koyanic root is a single syllable, either heavy or light, but with a strictly laryngeal nucleus, i.e. the allowed nuclei are {*Xə *Xə̄ *ə *ə̄ *əX *ə̄X *X}. Roots are notated with √, e.g. *√mx₃n "to grow, increase".

{*əx₁ *ə̄x₁ *əx₂ *ə̄x₂ *əx₃ *ə̄x₃} regularly alternate with {*i *ī *a *ā *u *ū} when unstressed. This alternation can clearly be seen in compounds using the root *√təx₂w "to talk": Note that this applies only to root nuclei, /əX/ and /ə̄X/ can occur unstressed elsewhere. For example, the stem *zə́x₁ɲ̥cʰəx₂ "berry" exhibits unstressed *əx₂.

Derivation
Word stems are typically formed by attaching an affix to a root or compounding roots. Not all words are formed from roots, some words are polysyllabic and contain other vowels but cannot be morphologically broken down, e.g. *tʰuyə̄ “house” and *ir̥luy “ball”. Compounds of two stems are common, with the typical patterns shown below.
 * Noun-transitive: A transitive verb with noun as object
 * (In)transitive-noun: A noun which performs a verb
 * (In)transitive-adjective: A verb characterised by an (adverbial) adjective
 * Noun-adjective: A noun characterised by an adjective

Infixation
Verb conjugation makes use of some infixes which are sensitive to phonotactics. The infix is inserted immediately after the first consonant of the base, or attached as a prefix on a vowel-initial base. An epenthetic *⟨ə⟩ may be inserted before the infix if phonotactically required. Some examples of infixation using hypothetical infixes on real reconstructed forms, and their surface phonological forms with prosodic structures, are shown below.


 * *⟨kʰ⟩ ???
 * *⟨nə⟩ ???
 * *⟨is⟩ ???

Root-based affixes
There are some derivational affixes used exclusively to derive stems from roots. These are generally prefixes and infixes.
 * *⟨qx₂əq⟩ derives transitive verbs from roots with typically intransitive meaning, e.g. *tə⟨qx₂ə́q⟩aw "to speak with, tell" from *√təx₂w "to speak".
 * *ə̄- derives intransitive verbs from roots with typically transitive meaning, often with an involuntary or passive meaning, e.g. *ə̄́-tʰə̄m "to sink, drown" from *√tʰə̄m "to submerge".

Derivational suffixes
There is also a separate category of derivational suffixes which can be applied to stems or roots to derive new stems, usually involving a change of word class. A selection of reconstructed suffixes are shown below.

Nouns
Proto-Koyanic had several cases depending on the animacy of the noun. The animate series of cases also marks plurality in the accusative, ergative, oblique and genitive cases.

Transitive Verbs
Transitive verb stems were inflected with a large set of prefixes, infixes and suffixes. Each major AQ language uses these affixes in slightly different constructions and patterns, meaning the overall system of conjugation at the time of Proto-Koyanic is not known. The morphemes which have been reconstructed are shown below, though how they interacted is not clear, so this table should not be assumed to represent individual forms of a single conjugation paradigm.

Intransitive Verbs
Intransitive verb stems have a similarly reconstructed set of inflectional affixes, shown below.

Word order
Notable features of Proto-Koyanic are the rich morphological system in form of affixes in both verbs and noun, featuring two genders, inanimate and animate nouns are labelled with asymmetric morphology, inanimate nouns are the least marked ones they do not feature plural numbers. As opposed, animate nouns are the most marked ones with three kind of genitives working together, and marking the singular absolutive. Proto-Koyanic word was flexible, but basic declarative sentences typically had the structure SVO.

Relative clauses
Relative clauses in Proto-Koyanic carries either zero morphology or it is used the particle *məx₂ŋ to express relative clauses. Languages

Pronouns
First and second person pronouns in Proto-Koyanic distinguished three cases, absolutive, genitive and instrumental; while the third person pronouns have additional ergative and dative marking.

Numerals
Proto-Koyanic, like all its descendants, used a base-10 system of numerals. The ten basic numerals are shown below:

Relation to Upic
The numerals 1-5 are thought to be related to those forms reconstructed for Proto-Upic, the common ancestor of the Upic languages. This relationship has proven useful in dating Proto-Upic and placing bounds on the time period of early Upic splits.

Alternative reconstructions
The form of Proto-Koyanic presented in this article is the most well established and commonly used, but not the only possible reconstruction. Various other reconstructions of the language may sometimes appear on other pages of the wiki or elsewhere, a few of which are briefly discussed below.

Original reconstruction
The first full reconstruction of Proto-Koyanic was largely the same as the one currently used, establishing most of the core features. At this time, the language was known as Proto-Arklobu-Qachkav, as a lack of data led to the conclusion that Arklobu and Qʼachkav were related only through the most recent common ancestor of all Koyanic languages. In fact, these languages have been shown to have a more recent common ancestor further down the phylogenetic tree, still known as Proto-Arklobu-Qachkav.

A few key differences between the original and current reconstructions are shown below:


 * The vowel qualities were slightly more speculative, suggesting *ɑ *ɑː *ʊ *ʊː for what are now usually given the less phonetically precise labels *a *ā *u *ū.
 * Less emphasis was put on the importance and role of stress in the formation of vowel qualities, now generally thought to be an integral part of Proto-Koyanic's phonology.
 * The nature of the alternation between the cardinal vowels and laryngeal nuclei was not as well understood.
 * Voiceless sonorants were considered completely phonemic and in slightly different distributions, leading to various more complex reconstructed forms which have been proven unnecessary by newer analysis. Despite this, the voiceless glides *ẙ *w̥ were initially not considered at all, and later appeared only rarely.

Late Proto-Koyanic
Late Proto-Koyanic (LPKn) was an attempt to reconstruct the ancestor to all Koyanic languages except the Duodoseic branch, a group sometimes called the Nuclear Koyanic languages. The motivation for this exercise was the undoubted fact that the vast majority of Koyanic languages reflect a common set of sound changes which remodelled Proto-Koyanic's laryngeal system into a large inventory of contrastive vowels. The approach taken was essentially an extension and reinterpretation of more traditional reconstructions of PKn, positing a few sound changes which brought it closer in line with Proto-Arklobu-Qachkav. Late Proto-Koyanic ultimately proved an unnecessary complication in understanding Koyanic development and is generally no longer used, but was incidentally of great benefit to an entirely novel reconstruction of true Proto-Koyanic, detailed further below.

Three characteristic changes involving the laryngeals *x₁ *x₂ *x₃ were assumed to have occurred concurrently, forming the basis of Nuclear Koyanic:
 * 1) Laryngeal colouring of schwas, generating pre-glottalised and post-glottalised vowels *ˀi *ˀī *ˀa *ˀā *ˀu *ˀū *iˀ *īˀ *aˀ *āˀ *uˀ *ūˀ
 * 2) Vocalisation of laryngeals into fully glottalised vowels *ĩ *ã *ũ
 * 3) Merger of laryngeals into a single consonant *x

These changes represent an example of transphonologization, where the contrastive feature which was carried by the laryngeal consonant in Proto-Koyanic is shifted to the coloured vowels, but no overall change in phonemic contrast has occurred. In contrast, PAQ is thought to have undergone structural changes to the phonology, developing glottalised obstruents and new vowel qualities, while lacking vowel phonation. The proposed LPKn changes might not actually represent any innovation or meaningful alteration to the phonemic contrasts present in the supposedly earlier Proto-Koyanic, and so they are perhaps more sensibly understood as a reanalysis of Proto-Koyanic more than an actual historical development. For this reason, Late Proto-Koyanic is no longer considered a canonical historical language stage, but went on to form the basis of a completely new reconstruction of Proto-Koyanic, detailed below.

The resulting remodelled phonology retains many features of Proto-Koyanic, but is structurally distinct. It was also suggested that LPKn may have featured the development of true palatals *c *cʰ from older palatalised velar consonants *kʲ *kʲʰ, an idea which eventually led to the novel reconstruction.

Novel Proto-Koyanic
The novel reconstruction, often called Novel Proto-Koyanic (NPKn) is intended to comprehensively explain as many irregularities and inconsistencies with previous reconstructions, while remaining a naturalistic description of a language. This reconstruction is based on the work done on Late Proto-Koyanic, and focuses more on features absent from Arklobu-Qʼachkav languages which are unique to earlier diverging groups like Duodoseic and Kát. Despite supposedly aiming to be as evidenced-based as possible, this reconstruction is much more speculative than traditional reconstructions, thus has many controversial elements which are yet to be widely accepted.

The phonology of the novel reconstruction features several drastic changes to the interpretation of Proto-Koyanic phonology:


 * Consonants are considered in two groups, central and peripheral, with each group having unique properties distinguishing it from the other group.
 * Plosives are reconstructed with an unspecified fortis-lenis distinction, which it has been suggested could have been voice rather than aspiration.
 * Palatals and sibilants are reinterpreted as palatalised plosives to explain a few odd reflexes and patterns observed in early Koyanic languages.
 * The laryngeal consonants *x₁ *x₂ *x₃ are rejected completely, with a rich system of vowel phonation reconstructed in their place.

Central consonants
The central consonants contrast three levels of sonority hierarchy, similarly to traditional reconstructions of Proto-Koyanic:


 * Obstruents — In the form of plosives with a variedness distinction, may only occur at syllable boundaries
 * Nasals — May additionally occur in medial positions, closer to the syllable nucleus than the obstruents
 * Approximants — May occur even closer to the nucleus than nasals, producing the familiar maximal syllable structure CNRVRNC.

The novel reconstruction then introduces palatalised forms of each of the eight central consonants. These are phonemically distinct single consonants which may occur in all the same positions as the plain forms. They should not be considered equivalent to sequences with medial *-y-, which NPKn supposes are forbidden for the central consonants.

The reconstruction of palatalised consonants may explain two phenomena unique to Duodoseic:


 * Reconstructing palatalised velars *gʲ *kʲ rather than true palatals *c *cʰ more naturally explains the Proto-Duodoseic reflexes as velar consonants *k *kʰ, since an unconditioned palatal > velar shift may not be possible.
 * Two PKn roots traditionally reconstructed *zwəx₁p "birch" and *swəkʰ "string" are reflected by Proto-Duodoseic *dwēp and *twōkʰ respectively. In NPKn, these roots are *dʲwe'b and *tʲwək respectively, with the Duodoseic reflexes explained as depalatalisation of *dʲ *tʲ to *d *t immediately preceding the opposing unpalatalised glide *w. In all other circumstances, *dʲ *tʲ would then shift to *z *s, as is observed in all daughter languages.

Peripheral consonants
Peripheral consonants had no palatalised forms, but could cluster with palatalised medials unlike the central consonants. The peripheral consonants exhibit no approximants, with the only peripheral sonorant being the labial nasal *m, which may occur in the same positions as the other nasals.

While *x₁ *x₂ *x₃ are rejected, the other laryngeal consonants *h *ʡ are reconstructed much the same as usual, patterning as a fortis-lenis obstruent pair.

Onset clusters
An additional constraint is that palatalised medial consonants *-nʲ- *-ŋʲ- *-lʲ- *-y- may only occur following peripheral consonants, while central consonants may only take unpalatalised medials. If a cluster of a central consonant and a following palatalised medial were to arise morphologically, it is believed a special rule would shift the palatalisation onto the initial consonant, e.g. *k-nʲ-, *ŋ-y- become *kʲn-, *ŋʲ- respectively. Palatalised consonants could still occur before a medial *-w-, an important consideration in their reconstruction.

This theory simply explains the lack of medial *-y- following the traditional PKn consonants *t *tʰ *z *s *k *kʰ *c *cʰ *ɲ *r *l, as well as the more curious lack of the clusters *tr- *tʰr- *zr- *sr- *kr- *kʰr- *cr- *cʰr-, while their equivalents with *-l- are abundant. NPKn proposes that *r was originally palatalised *lʲ, thus being forbidden from clustering with other central consonants. If these forbidden clusters arose morphologically, they would shift the palatalisation to the onset, yielding clusters which are commonly attested:


 * {*d-lʲ-, *dʲ-lʲ-} → *dʲl- (traditional *zl-)
 * {*t-lʲ-, *tʲ-lʲ-} → *tʲl- (traditional *sl-)
 * {*g-lʲ-, *gʲ-lʲ-} → *gʲl- (traditional *cl-)
 * {*k-lʲ-, *kʲ-lʲ-} → *kʲl- (traditional *cʰl-)

The traditional constraints regarding medial nasals in onsets also hold, that homorganic clusters are forbidden and nasal-nasal onsets are allowed only when one of those nasals is *m. Novel Proto-Koyanic was initially challenged on its failure to explain the clusters *zn- and *sn-, which are undeniably attested and would be reconstructed *dʲn- *tʲn- in NPKn, violating the well-established homorganic plosive-nasal rule. However, the lack of *cn- *cʰn- in traditional PKn means that these can actually be successfully reconstructed as *gʲn- *kʲn- in NPKn, with later development to *zn- *sn- by simple assimilation.

A table of all possible onset clusters of two consonants in NPKn is shown below, with the attested forms marked with a tick (✓).

Vowels

 * Modal voice — larynx remain open for the full duration of the vowel
 * Pre-laryngealised — initial laryngeal constriction, opens up through the duration of the vowel
 * Post-laryngealised — initially open larynx, closing into laryngeal constriction at the end of the vowel
 * Creaky — larynx remains constricted for the full duration of the vowel