Möxali

Möxali (also known as Möxali) is the primary language of the Lavondu people of T'ugü, and a lingua franca throughout surrounding coastal regions of Fountainhead, particularly those connected to T'ugü by maritime trade. Möxali emerged as a mixed language in bilingual communities of Xiyeru in T'ugü following the Xiyeru Exodus.

Phonology
Möxali is most commonly spoken as a second language, by many peoples over a wide area throughout its long history, so there is much variation in the phonology. There are, however, some patterns and general tendencies common to all speakers or particular subgroups. In order to view Möxali as a whole, its phonology is discussed here using the romanisation of its core phonemic inventory, which is common to almost all varieties of the language. Each phoneme can be understood as representing a diaphoneme, an abstracted phonological unit with an analogous phoneme in each variety of Möxali, which may have different phonetic realisations depending on the dialect or historical stage. Particular phonetic varieties are also given in square brackets where relevant.

Alveolar tap
The consonant r is an alveolar tap [ɾ] or occasionally a trill [r] which only occurs intervocalically, most commonly as an allophone of d. The origin of r lies in the intervocalic allophone of the Xiri rhotic [ɾ], while in initial position this consonant was pronounced [ɹ] and became Möxali l. Because of this, the vast majority of r stem directly from Xiri, though some words replaced intervocalic r with l (e.g. valon "yam wine", from Xiri warań) via early dialectal intermixing during the formation of Möxali.

In the native writing of Möxali, the Opsíbdañox, r is written with the same letters as l, and in formal texts l can be distinguished by adding the vikvínon which more typically functions as a devoicing diacritic. Despite it being taught as a separate consonant by the scribal schools in T'ugü, its phonemic status is marginal at best in even the most conservative Möxali, while in all Outer Möxali dialects r and intervocalic d are merged into [ɾ~r] as a rule. Educated speakers of Inner Möxali, particularly in its early literary history in the 9th and 10th centuries AGS, often prescribed the pronunciation of intervocalic d as [d̪], maintaining a distinction with r [ɾ] in slow, careful speech. By around the 14th century, all speakers of Möxali had merged r with intervocalic d, though the orthographic standards retain the spelling of r as separate to d.

Glottal stop
The glottal stop q is phonemic only in Inner Möxali, the dialect which remained spoken in T'ugü after the Möxali language was carried elsewhere by population migration. The group of emigrant dialects, Outer Möxali, do not have the phoneme q, but still have a phonetic glottal stop with predictable epenthetic placement between vowels. The phonetic glottal stop is romanised as ⟨ʼ⟩ for clarity, but is not represented in the native script. Compare the following examples: Note that common Outer Möxali spelling additionally does not use certain diacritics, thus l and s are not distinguished from r and z respectively.

Front-back allophones
Some consonants have a pair of allophones determined by the presence of a front (e ö i í) or back (a o u) vowel in the host syllable. These allophonic pairs or some variants of them are present in the vast majority of speakers, particularly those who speak Möxali as a first language or are in regular contact with those who do. The typical phonetic values of these allophonic pairs are shown below:

Vowels
The mid vowels (i u e ö o) are realised as [ɪ ʊ e̞ ø̞ o̞] in T'ugü and most typically [e o ɛ œ ɔ] elsewhere. ö in particular can vary in height quite freely between [œ] and [ʏ] depending on speaker and phonological environment, but the height hierarchy e < ö < i is usually maintained.

Phonotactics
Möxali's syllable structure is (OR/C)V(D), where O is an obstruent (p t c k b d f s x v z j), R is an approximant (v l y) with a different place of articulation to O, C is any consonant, V is any vowel (e ö i í a o u), and D may be any consonant except y (p t c k b d f s x h v z j l). Later Möxali dialects developed a further restriction on coda consonants which meant plosives were no longer allowed, described in more detail in the Morphophonology section below. Coda n nasalises the preceding vowel and, if there is a following consonant, manifests as a nasal stop which phonetically assimilates to the place of articulation of that consonant. since v patterns both as a fricative and an approximant in the context of syllable structure, it is the only consonant that can occur as a single onset, be either consonant in an onset cluster or be a syllable coda. The allowed syllable onsets are shown in the table below. Those shown in red are merged, single-consonant forms that surface when forbidden clusters are forced together. For example, |-h-y-| surfaces as - x -.

Morphophonology
It is useful to analyse the structure of Möxali as having an underspecified morphophonemic level, with phonotactics and other regular processes yielding the phonemic level. Morphophonemic forms are shown in |vertical bars|. In addition to the phonotactic constraints on onset clusters described above, there are five main phenomena; coda neutralisation, obstruent devoicing, inversion, vowel sandhi and vowel elision.

Coda neutralisation
An underlying stop consonant in morpheme-final position is subject to neutralisation if it arises as a syllable coda at the phonemic level, i.e. if there is no following syllable in which the consonant can act as the onset.

Nasals neutralise into assimilatory coda n:


 * ösem-∅| → ösen (firewood-ᴀʙꜱ)
 * ösem-in| → ösemin (firewood-ᴇʀɢ)

The donation of a stop to a following syllable can occur across morpheme or word boundaries, meaning the phonological form of a word can depend on the following word:ösem-∅ í-∅-bílbil

firewood-ABS IMP-PV-gather.up

"Gather up some firewood!"While coda plosives were initially allowed in Möxali and are retained in the Inner dialect, they were lost in the later and more widespread Outer dialects. In these dialects, plosives undergo spirantisation when they fall in coda position, in which they mutate to the corresponding fricative with the same voicedness and place of articulation: p t c k b d → f s x h v z. This results in inflectional alternation like the following:


 * límok-∅| → límoh (bee-ᴀʙꜱ)

Despite this spirantisation rule being present in the most commonly spoken forms of Möxali throughout history, the native orthography tends to favour more conservative spelling in formal contexts, e.g. ⟨LÍ-MO-KA⟩ for límoh. For consistency and clarity, romanisation of Möxali generally uses conservative spelling unless otherwise stated, i.e. límok not límoh.
 * |límok-en| → límokin (bee-ᴇʀɢ)

Obstruent devoicing
Voiced obstruents (b d j z v) devoice to (p t c s f) when in a cluster with any voiceless consonant (p t c k q f s x h). For example, the Alöbi verb ubsöb (write) is reflected in Möxali as upsöb, the adjunct form of which opsíb-d-a (written) is the source of the name of Möxali's script opsíbdañox "sacred written thing". The stativiser suffix -d will also similarly devoice, for example the verb mídös (to smoke, as of meat) yields the adjunct form midista (smoked). This voicedness alternation is most commonly seen in the aspect conjugation of verb roots: The exception to this rule is the consonant v, which acts as a medial approximant when following another consonant, thus is not itself subject to obstruent devoicing. For example, the Alöbi adjective wik'wa (different) is reflected in Möxali by víkva [ˈβi.kβ̞ɑ] rather than *víkfa [ˈβik.ɸɑ].

The same devoicing rule applies across word boundaries, as well as at the boundaries of an utterance, though b often remains voiced in the latter case. For example, the phrase den-díhaj spoken in isolation would typically be rendered as [t̪ˢɛn̪d̪ᶻiχɑtʃ].

Inversion
Inversion is an ablaut process which acts on both vowels in a disyllabic stem. It developed out of an archaic stress rule in Xiri which is no longer active in Möxali, and through analogy has become a regular process triggered by certain suffixes. Vowels i í u ú in the first syllable are lowered to e i o u, and vowels e ö i o u in the second syllable are raised to i i í u ú respectively:


 * |cílom-∅| → cílom (blood-ᴀʙꜱ)
 * |cílom\⇵-yuvek| → ciluñuvek (bloodfish)

When a polysyllabic word is formed by attaching morphemes to the front of the base stem in this way, the stem itself still undergoes inversion in its inflection:


 * cílom\⇵-yuvek-∅| → ciluñuvek (bloodfish-ᴀʙꜱ)
 * cílom\⇵-yuvek\⇵-ä| → ciluñovike (bloodfish-ᴀᴄᴄ)

Exactly what constitutes the Möxali stem is difficult to rigidly define, some polymorphemic structures count as stems while others do not, but stems are always disyllabic. For example, the additive numerals, which are special forms of the basic numeral roots which were formed in Alöbi with the addition of a suffix -ö. In Möxali, these additive numerals exist only in compounds and are considered disyllabic stems, so are always inverted:


 * |uj-| uj- (one) → |ujö\⇵-kvöj-| ojikvöj- (fifty one)
 * víb-| víb- (two) → |víbö\⇵-kvöj-| vibikvöj- (fifty two)
 * níf-| níf- (three) → |nífö\⇵-kvöj-| nifikvöj- (fifty three)
 * bíx-| bíx- (four) → |bíxö\⇵-kvöj-| bixikvöj- (fifty four)

Vowel sandhi
Morphemes with an initial vowel avoid placing that vowel adjacent to a previous vowel by a sandhi process determined by the qualities of the two vowels. In the context of sandhi, (e i í) and (o u ú) form natural groups labelled E and O respectively.

If the first vowel is higher than the second vowel on the hierarchy a < {e ö o} < {i í u ú}, an epenthetic approximant is inserted between the two vowels, y after front vowels and v after back rounded vowels: The vowels in brackets are subject to regular elision as described in the next section.

If the first vowel is the same height or lower than the second, they merge into a single vowel with quality determined by the following table: A full table of possible vowel sandhi combinations is shown below: The close vowels (í ú) are considered equivalent to the mid-close vowels (i u) in sandhi processes. This equivalence has two consequences:


 * If a merged vowel i or u lands in a stressed syllable, which it most often does, the resulting quality will be í or ú respectively.
 * í and ú are not considered higher than i and u, thus |í-i| does not yield íye, but í. Note that i, and by extension í, are both considered higher than e, so |i-e| yields iye.

In Outer Möxali (OM) dialects, the lack of a glottal stop at the phonemic level can obscure the predictable behaviour of the vowel sandhi. Words which historically began with the glottal stop q, which is retained in Inner Möxali (IM), do not undergo vowel sandhi despite being apparently vowel-initial. These can be analysed as having an underlying glottal stop, which technically need not exist at the phonemic level despite sometimes manifesting at the phonetic level. Compare the following examples:
 * |ʼañon| añon (lid, cf. IM qañon) → |xu-ʼañon| xuʼañon (ᴘʟ-lid), underlying glottal stop blocks interaction between vowel qualities
 * |onxub| onxub (adze, cf. IM onxub) → |xu-onxub| xvonxub (ᴘʟ-adze), sandhi interaction present

Note that these processes apply at morpheme boundaries, thus the full range of Möxali’s phonology can be expressed elsewhere within the morpheme, e.g. |mua| manifests as mu’a, not *muva.

Vowel elision
Stressed vowels may never be elided, but there are several circumstances which trigger the regular elision of unstressed vowels. One such circumstance is in the consonant epenthesis of vowel-initial morphemes described above:


 * su-azi| → *s(o)ˈvazi → ˈsvazi

This elision is blocked by an approximant in the syllable onset:


 * |kyo-azi| → *ky(o)ˈvazi → kyoˈvazi

If the approximant has the same place of articulation as that preceding onset, it too is elided:

Some morphemes, particularly short inflectional affixes, have two allomorphs, one with a and one where the vowel is elided. This is shown in square brackets in the underlying morphophonemic notation. The presence of the vowel depends on its morphophonological environment and functions as epenthesis wherever phonotactically required. For example, consider the active voice prefix |f[a]-|:
 * ijí-azi| → *íj(i)ˈyazi → *iˈj(y)azi → iˈjazi (Note also the inversion of the initial stem íji- → ijí-)


 * When affixed to vowel or approximant-initial stem, the [a] is not necessary and so is not present:
 * f[a]-u-qa| → fuqa (ᴀᴠ-eat-ᴘꜰᴠ)
 * f[a]-la-qa| → flaqa (ᴀᴠ-cook-ᴘꜰᴠ)
 * When affixed to other stems, the [a] is required to break up the fC- onset cluster:
 * f[a]-mum-qa| → famunpa (ᴀᴠ-fill-ᴘꜰᴠ)
 * |f[a]-jí-qa| → fajíqa (ᴀᴠ-bite-ᴘꜰᴠ)

Dialects
The history of Möxali features two major dialect groups, Inner and Outer Möxali, which each have unique features and geographic distributions.

Inner Möxali
Inner Möxali was spoken primarily as the first language of the Lavondu in T'ugü, thus its evolution was shaped by continued contact with Alöbi. It was generally the most conservative dialect, featuring comparatively little change from the earliest attested Möxali c. 8th century AGS. This conservatism partly stems from the origins of Inner Möxali as a register of the early spoken language, which may have originated from a deliberate effort by the Lavondu to preserve a more "authentic" manner of speech. The early evolution of the speech patterns and vocabulary of this register were primarily guided by historical texts, resulting in many unique features produced by a what was essentially feedback loop. Newly written texts aiming to emulate the styles of older texts often introduced new quirks which informed the speech of later speakers.

The orthographic tendencies of all variants of Möxali are continuations of earlier forms of the language, meaning they most closely resemble the conservative dialect. Despite this association, Inner Möxali should not be confused with a literary language, of which there was no clear standard for most of Möxali's history.

The most characteristic feature of Inner Möxali phonology is the retention of coda plosives, which are not present in the outer dialects. The other unique features are mostly superficial differences in the pronunciation of certain phonemes:


 * the labial fricatives (f v) are bilabial [ɸ β], rather than labiodental [f v]
 * j is typically pronounced as an affricate [dʒ~dʑ], rather than a fricative [ʒ~ʑ]
 * the vowels (i ö u) are realised as [ɪ œ ʊ], rather than close-mid [e ø o]

The front/back consonant allophones described above were present in first language speakers, but less so in second language speakers, i.e. Öb who spoke Alöbi as a first language. In Öb-style Inner Möxali, the coronal obstruents (t d s z c x j) were [t̪ d̪ θ ð tʃ ʃ (d)ʒ] in all environments, never the allophones [t̪ˢ d̪ᶻ s̪ z̪ tɕ ɕ (d)ʑ] which were present next to front vowels for most other speakers.

Outer Möxali
Outer Möxali was a collection of dialects which were spread throughout the coast of Fountainhead beginning around 900 AGS. There was more variation in the outer dialects than in Inner Möxali, but there still existed many common features which distinguished them as a related group. Outer Möxali was spoken as a first language by some Lavondu colonies along Fountainhead's coast, but most commonly as a second language by many other cultures, including but not limited to: the Xiyeru, Ansang peoples, and speakers of Nambāno dialects.

Whereas Inner Möxali adopted many unique monomorphemic verb stems from Alöbi, the outer dialects used far fewer, instead tending to innovate new stems more readily using root-aspect constructions.

All outer dialects exhibit the coda spirantisation rule described in the morphophonology section above, leaving no coda plosives at the phonemic level and inflectional alternation between fricatives and plosives in many words. For example, compare the inflectional set tanih/tanikin/tanikeb "knife" (with underlying |k| to ínih/ínihin/íniheb "finger" (with underlying |h|). The consonants (f v) are almost always labiodental [f v] in the outer dialects, but /v/ was always a bilabial approximant [β̞] when acting as the medial approximant in a syllable onset cluster. The dental consonants (n t d s z l) are often alveolar [n t d s z l] rather than true dental [n̪ t̪ d̪ s̪~θ z̪~ð l̪], or in some regions these are in free variation.

Another common feature of Outer Möxali is the rhotacism of d to [ɾ~r] when following a vowel, i.e. between vowels and in coda position. This rhotacism generally does not represent a phonemic change, though some dialects later developed a distinction between /r/ and /d/, and such cases of /d/ [ɾ~r] are sometimes written as ⟨R⟩ rather than ⟨D⟩ even in those dialects that did not.

Written language used conservative spelling in formal and culturally important texts, perhaps due to the cultural influence of Inner Möxali or just a continuation of older tradition, while more everyday writing frequently featured unique spellings which more accurately represented dialectal forms. The letters ⟨QA QE⟩ were dropped completely in written Outer Möxali, reflecting the loss of the phonemic glottal stop q. The glottal stop [ʔ] still occurs at the phonetic level as epenthesis between two adjacent vowels, blocking sandhi in some circumstances. This can be analysed as an underlying glottal stop, but it is not necessary to consider it a separate consonant, which is likely why the Q letters fell out of use in favour of the null-onset vowels. Despite lack of representation in the native script, the phonetic glottal stop in Outer Möxali is romanised as ⟨’⟩ within words for clarity of reading.

Grammar
Möxali is an agglutinative language, in which words are formed by attaching several classes of affixes to disyllabic word stems. Transitive verbs, statives and numerals have their stems derived from monosyllabic roots with some classificatory suffix attached, while noun stems are disyllabic morphemes which cannot be broken down.

Transitive verbs take inflectional affixes for mood, voice and aspect but are not inflected for person, number or tense. Tense is generally denoted by temporal adjuncts or understood from context. The person of the subject may optionally be marked with a prefix to refer back to a previously established subject, but this is not strictly part of the verb's inflection. Stative verbs take a single inflectional suffix for the grammatical number of their subject.

Möxali's morphosyntactic alignment has characteristics of ergativity and its transitive verbs use a system of symmetrical voice to mark the syntactic role of the subject of the verb. Nouns are marked for absolutive, ergative and accusative case and have additional optional number marking.