Proto-Namb language

Proto-Northern Gäj (abbreviated PNGäj) is the reconstructed ancestor of the Nambāno and Arklobu languages, which together form the Northern branch of the "Gäj" languages.

*ɸ and *θ
The fricatives *ɸ and *θ developed from Proto-Gäj *pʰ and *tʰ before stops (nasals and plosives). Although these consonants are reconstructible in PNGäj, their phonemic status is not clear, since their distribution is complicated by stem-mutation, an apophonic process characteristic of the Gäj languages. Instances of *ɸ and *θ are generally present only in inflectional forms where a consonant-initial suffix was attached to a stem-final *pʰ or *tʰ. For example, Proto-Gäj *pʰɛ́pʰ-mi, *ǣ́tʰ-mi "leaf-ɢᴇɴ, snake-ɢᴇɴ" become Proto-Northern Gäj *pʰə́ɸ-mi, *éθ-mi. The difficulty in understanding whether these consonants are phonemic comes in certain inflectional suffixes which triggered fronting of the stem vowel but did not have an initial consonant, for example the PNGäj reflex of PGäj *pʰɛ́pʰ-īr "leaf-ᴇʀɢ" may have been either *pʰə́pʰ-yɨr or *pʰə́ɸ-yɨr, with phonemic /ɸ/. A more detailed examination is presented in the section on stem mutation below.

Semivowels
The semivowels *y and *w occurred only as medial consonants in onset clusters, having historically been fortified to *ɟ and *gʷ respectively in onset position and lost in medial and coda position, affecting the quality of the preceding vowel in the latter case. Semivowels were rare in Proto-Gäj but became more frequent in Proto-Northern Gäj due the vowel shift *ī *ū *ē *ō > *yɨ *wɨ *yə *wə.

*ö and *ü
Unlike *ä which was a fully contrastive vowel, the front rounded vowels *ö and *ü may not have been phonemic in Proto-Northern Gäj. These fronted forms arose as allophones of *o and *u before *i in the next syllable, an environment commonly associated with stem-mutation, thus their phonemic status depends on the extent of analogical levelling of stems in PNGäj, much like *ɸ and *θ. If *ö and *ü were extended to forms without *i in the next syllable, this would remove them from the allophonic trigger environment and therefore yield a phonemic contrast with *o and *u.

Stem mutation
Proto-Gäj featured a pattern of grammatical alternation where stem-final consonants in certain inflections would mutate according to the rule:


 * *b *-d *-ɟ *-g *-r > *-p *-t *-c *-k *-s

In Northern Gäj, this developed further due to two sound changes affecting the mutated stem. The first spirantised aspirated plosives before stops, while the second shifted the back vowels *a *o *u to front vowels *ä *ö *ü, triggered by the presence of *i or *ī in the following syllable. Additionally, *s was lost, yielding the final set of mutation rules:


 * *b *-d *-ɟ *-g *-gʷ *-r > *-p *-t *-c *-k *-kʷ *-∅
 * *-pʰ *-tʰ *-cʰ *-kʰ *-kʷʰ > *-ɸ *-θ *-ç *-x *-xʷ
 * *-a- *-ə- *-ɨ- *-o- *-u- > *-ä- *-e- *-i- *-ö- *-ü-

Stem mutation is the result of coincidental inflectional morphology causing multiple independent sound changes to frequently occur together, and the compound phonological variation which arose becoming recognised as a common pattern. This separated a stem into two distinct forms strengthened by semantic and morphological analogy. For example, the first and second person imperfective suffixes *-cʰɨr (< *kʰīkʰ) and *-cʰɨrkʰ (< *kʰī⟨r⟩kʰ) triggered ablaut of the previous vowel in the stem and blocked the voicing of a stem-final plain plosive which had become voiced in the perfective and irrealis conjugations, or triggered spirantisation of an aspirated plosive. In nominal inflection, the genitive suffixes *-mi, *-ɲɨŋ (< *-mīŋ) also triggered the same ablaut and consonant mutation, as did the ablative suffixes *-çɨl (< *-çīl) and *-cʰɨl (< *-tʰīl).

Many paradigms of related forms which partially or inconsistently mutated the stem were analogically levelled into systems based around a singular mutated stem. The semantic relationships between forms may also have influenced this analogy, for example in the nominal cases. There may have been a semantic split into more active or dominant roles like the possessive, genitive and ablative which happened to use the mutated stem, and passive/recipient roles like the absolutive, accusative and lative cases using the basic stem. In such a system, the ergative case would likely be more associated with the active roles, and this may be part of what led to Proto-Northern Gäj using the mutated stem for the ergative case, unlike the other Gäj languages.

Mutated ergative
The fronted vowel became the most important characteristic feature of the mutated stem in PNGäj, evidenced by other morphological changes revolving around the ablaut. For example, consider the inanimate ergative suffix *-īr, which triggered voicing of the stem-final obstruent in Proto-Gäj, yielding forms like *ā́g-īr and *pwób-īr from PKn *ā́q-īr "hill-ᴇʀɢ" and *pwə́p-īr "moss-ᴇʀɢ" respectively. In Proto-Northern Gäj, these underwent vowel fronting, so should have yielded the expected forms *éz-ɨr and *kʷőb-yɨr respectively, but instead the forms *é-ɨr and *kʷőp-yɨr are reconstructed. This indicates that these forms were changed by analogy with other mutated stem forms which usually front the vowel and devoice the final consonant together, i.e. *é- and *kʷőp-. Barring a few irregular exceptions, this analogical step firmly placed the ergative singular into the mutated stem category in the Northern-Gäj languages, distinguishing them from other Gäj languages which use the basic stem for the ergative case.

The ergative plural suffixes *-ä and *-ond exhibit neither of the features which trigger the two mutations, yet may also have attached to the mutated stem. There is some evidence to suggest that the analogy was extended this far in PNGäj, but it is difficult to reconstruct. If this did occur, this would have yielded front rounded vowels *ö and *ü in environments in which they would not have arisen allophonically, thus would contrast with *o and *u respectively. Compare *xúr-e "bird-ᴀʙꜱ.ᴘʟ" with **xű-ond "bird-ᴇʀɢ.ᴘʟ", which could have been formed by analogy with *xű-mi "bird-ɢᴇɴ.ᴘʟ" and *xű-yɨl "bird-ᴀʙʟ.ᴘʟ". Similarly, the genitive singular *-mäx may have appeared on a fully mutated stem with the fronted vowel.

Phonemicity of *ɸ and *θ
A problem arises in the reconstruction of stems with final *pʰ and *tʰ, which according to the aspirated plosive mutation rule would yield *ɸ and *θ respectively in the mutated stem. Unlike the dorsal fricatives *ç *x *xʷ which pose no issue, the fricatives *ɸ and *θ were not otherwise phonemic and would have been made phonemic by applying the mutation with originally non-mutating suffixes.

For example, PGäj *pʰɛ́pʰ-īr "leaf-ᴇʀɢ" would be expected to result in PNGäj *pʰə́pʰ-yɨr according to regular sound changes, but mutation by analogy with the genitive *pʰə́ɸ-ɲɨŋ and ablative *pʰə́ɸ-cʰɨl may have produced *pʰə́ɸ-yɨr. This could have yielded *ɸ in a phonological environment where it would contrast with *pʰ, though no such contrasts are reconstructible. Nambāno is known to represent the result of such analogy, for example PGäj *pʰɛ́pʰ-īr, *pʰɛ́pʰ-mīŋ, *pʰɛ́pʰ-tʰīl are reflected by Pre-Nambāno *pʰéɸ-yɨr, *pʰéɸ-ɲɨŋ, *pʰéɸ-cʰɨl. The only other daughter branch of Proto-Northern Gäj, Arklobu, simply merges the aspirated plosives into fricatives in all positions, yielding a fricative from an ancestral aspirated plosive in both stem forms regardless of mutation, for example Pre-Arklobu *ɸéɸ-yɨr, *ɸéɸ-ɲɨŋ, *ɸéɸ-çɨl. Despite both branches reflecting fricatives in the ergative (*pʰéɸ-yɨr vs. *ɸéɸ-yɨr), PNGäj may still have reflected *pʰ, since the analogical step may have taken place only in Nambāno later on, changing *pʰépʰ-yɨr to *pʰéɸ-yɨr by analogy with *pʰéɸ-ɲɨŋ and *pʰéɸ-cʰɨl.

The phonemic status of *ɸ and *θ is therefore technically not reconstructible, though it may be reasonable to assume all these analogical processes took place around the same time, which would have yielded phonemic *ɸ and *θ in PNGäj. By convention and for simplicity, *ɸ and *θ are usually transcribed as separate consonants in Proto-Northern Gäj, including as mutated forms of *pʰ and *tʰ.

Some examples of mutated stems which have been analogically levelled are shown in the table below.