Mila

Introduction

Mila is the language spoken by the lula on the planet Kukca. The lula come originally from a technologically advanced planet. The planet Kukca had been terraformed many million years previously and was deemed ripe for colonisation. The colonists were sent under the pretext that they would be repairing the on-planet communication system (in the form of a pyramid), but were not told that they would subsequently become colonists, (it has been discovered that finding willing volunteers is a far from easy task and the home planet necessarily resorted to tricks to send colonists). The colonists had been genetically altered to thrive on the planet, based on readings sent from the communication towers.

Other planets which have been previously terraformed have become 'zoos' which may be occasionally visited (our own Earth is one such planet), holiday destinations or like this one colonies, and partly as an elaborate social experiment from the home planet to see how lula can cope. The colonists first arrived about 150 years previous to the language as described here is spoken. On colonisation the language was completely regular having been the constructed lingua franca of the home planet.

Goals

The original aim of Mila was not to create a whole language fitting into a constructed world, but rather a way to experiment with language change. I wanted to create a language with a simple phonological system and phonotactics which may be manipulated to explore how language can change and to test certain linguistic hypotheses, historical phonology being my main area of interest. I soon discovered though that it would be a good chance to improve my knowledge of other areas of linguistics and it has inevitably developed into a full language. I didn't want at an early stage to deal with loan words which is why I felt it necessary to have the language in complete isolation and a constructed world has thus also been created. With a constructed world and culture it brings with it all the problems of translations from Earth-based languages. How does one translate 'dog' for example? Mila does have a word for 'dog' but it should be understood as a dog-like creature, as opposed to an actual dog. As such the goals of the language have somewhat drifted, now focussing on a realistic setting for the language, though development of the language into separate daughter languages is also an ultimate goal.

Design

As mentioned the sound system is very simple consisting of 11 consonants and two (or three) vowels depending on the phonological analysis. Commonly found consonants have been chosen, with the possible exception of palatal stops. There is a high and a low vowel, the high one being allophonically predictable based on the following consonant, although there are irregularities in both the consonant and vowel systems because of loan-developments from separate dialects and phonological processes and shortenings. I've tried to push myself with the grammar to make it as un-European as I could, but it is after all a language and as such will have certain similarities to other languages, although these similarities were not a conscious decision.

History

The lula who first inhabited the planet were taught raised by scientists in isolation, a necessity as they had been genetically moderated to cope with the planets climate (which is much hotter than the home planet) and air (which is richer in oxygen than the home planet). The scientists, being from an international background, spoke the planet's lingua franca, which is itself a constructed language, based on many of the home planet's language families. The original Mila (meaning simply 'language') was therefore completely regular. In the 150 years since colonisation the original population of 56 has risen to a few thousand (they breed quickly) and the language is now the everyday mother-tongue of the speakers, a situation which has never occurred on the home planet, such that Mila on the home planet has changed little in years. As a mother-tongue, though, the language has undergone a number of changes and irregularities have crept in, most notably in the phonology, shortenings of pronouns and semantic shifts.

Phonology

The Mila phonological system consists of 11, 14 or 15 consonants and two or three vowels, depending on the analysis as shall be described below.

Consonants

	Bilabial	Alveolar	Palatal	Velar	Glottal
Plosive	p   (b)	t   (d)	c   (ɟ)	k   (ɡ)
Nasal	m	n
Fricative		s   (z)			h
Trill		r
Approximant			j
Lateral (velarised)		ɫ

Notes on the consonants:

Voiceless Plosives

/p/ This is a bilabial voiceless stop [p]. It is unaspirated with voicing occurring very shortly after its release. It may occur in onset and coda position. In coda position before a following consonant it is unreleased with the release occurring on the following onset consonant. Phrase finally it is usually unreleased but this varies by speaker. Examples: oωɾ /ʹpinda/ [ʹpindɐ] many; coo /ʹsupa/ [ʹsupɐ] same; ooɔ /ʹpupka/ [ʹpup˺ka] sick; υυó /luʹlap/ [ɫuʹɫap˺ ~ ɫuʹɫap] I.3.

/t/ This is an alveolar voiceless stop [t]. There is some speaker variation as to its exact production. Women generally produce this sound apically whereas men have a tendency to produce this laminally, [t̻]. When it is produced laminally is may be slightly affricated as [t̻s] in pre-stressed position though affrication does not occur elsewhere. It is unaspirated with voicing occurring very shortly after its release. It may occur in onset and coda position. In coda position before a following consonant it is unreleased with the release occurring on the following onset consonant. Phrase finally it is usually unreleased but this varies by speaker. Examples: ɾɾυ /ʹtidla/ [ʹtidɫɐ ~ ʹt̻idɫɐ ~ ʹt̻^sidɫɐ] coin; oɾɾ /ʹpita/ [ʹpitɐ ~ ʹpit̻a] to crave; oɾɔ /ʹpitka/ [ʹpit˺kɐ ~ ʹpit̻˺kɐ] already; ɾ́ɾɾ /at/ [ɐt˺ ~ ɐt̻˺ ~ ɐt ~ ɐt̻] it.present.1.

/c/ This is generally described as a palatal voiceless stop [c] though there is allophonic variation as well as speaker variation. Allophonically this becomes [c̠] when followed by a non-front vowel, so when followed by /a/ or /u/. Many speakers also slightly affricate this sound to [cç ~ c̠ç̠] in pre-stress position. The pronunciation [ç] may also be heard in any position but is generally stigmatised. It is unaspirated with voicing occurring very shortly after its release. It may occur in onset and coda position. In coda position before a following consonant it is unreleased with the release occurring on the following onset consonant. Phrase finally it is usually unreleased but this varies by speaker. Examples: nɾʌ /ʹcidra/ [cidɾɐ ~ cçidɾɐ ~ çidɾɐ] red; nιo /ʹcuhpa/ [c̠uhpɐ ~ c̠çuhpɐ ~ çuhpa] big; ιnn /ʹhica/ [ʹhic̠ɐ] late; υnn:n /ʹlicic/ [ʹɫicic˺ ~ ɫicic ~ ʹɫiçiç] baby-sub; υnɾ /ʹlicta/ [ʹɫic˺tɐ ~ ʹɫic˺t̻ɐ] arm.

/k/ This is generally described as a velar voiceless stop [k] though there is allophonic variation. This sound becomes [k̟] allophonically before the front vowel /i/ and as such it forms a near merger with [c̠] mentioned above, though the two sounds occur in different environments and are recoverable if the following vowel is altered. It may occur in onset and coda position. In coda position before a following consonant it is unreleased with the release occurring on the following onset consonant. Phrase finally it is usually unreleased but this varies by speaker. Examples: ɔʌυ /ʹkila/ [ʹk̟iɫɐ] abyss; ɔυɔ /ʹkuga/ [ʹkugɐ] wall; ɔɔɔ /ʹkuka/ [kukɐ] earth; ɔɔɔ:ω /ʹkukin/ [ʹkuk̟in] earth-adj; ɔιɔ /kuʹhak/ [kuʹhak˺ ~ kuʹhak] gold.3.

Voiced Plosives

The voiced plosives are only marginally phonemic. In the earliest forms of Mila if a voiceless plosive and /s/ is preceded and followed by voiced sounds within the same word then the plosive will become voiced. As such voiced plosives were formerly only allophones of the voiceless plosives. Some language changes have occurred such that pronouns (and in one case a suffix) are shortened in common speech which has meant that voiced plosives now occur word initially in a limited number of words and [d] may occur word finally in words with the suffix :υɾ [-ud]. Also degemination has meant that former voiceless geminates are now simplex voiceless sounds which appear intervocalically, so in a position which previously triggered voicing. Their distribution is however very limited so can at best only be described as marginally phonemic. The Mila orthography does not have separate symbols for these sounds although the Romanisation does indicate them. Mila scholars therefore enjoy debating the status of these sounds, with purists denying their existence and looking at the phonology from a historic perspective, with progressives accepting them.

/b/ This is the voiced equivalent of /p/. It is a voiced bilabial stop [b]. It occurs marginally in initial position in unstressed shortened versions of some pronouns and is also possible in loan words which begin with a voiced bilabial. In this position voicing starts prior to the stop closure and manifests itself as prenasalisation. Its usual position is intervocalically or in a voiced environment and as such historically was an allophone of /p/ but degemination and shortenings have allowed for its phonemic status. It never occurs in word final position. Examples: ooó /ba/ [^mbɐ] it.absent.3; υoʌ /ʹlubra/ [ʹɫubɾɐ] enemy; υυo /ʹluba/ [ʹɫubɐ] I.

/d/ This is a voiced alveolar stop [d]. It does not display the speaker variation present for /t/ so may not be produced laminally and it also therefore resists affrication. As it has historically evolved from an allophone of /t/ its environments are more restricted than the voiceless stops. It may occur initially in unstressed shortened versions of some pronouns and is also possible in loan word which begin with a voiced alveolar. In this position voicing starts prior to the stop closure and manifests itself as prenasalisation. Its usual position is intervocalically or in a voiced environment. It may also occur in coda position in the isolated ending :υɾ /-ud/ doer. In coda position it is usually unreleased but this varies by speaker. Examples: ɾɾɾ́ /da/ [ⁿdɐ] it.present.3; υɾω /ʹlidna/ [ʹɫidnɐ] group; υʌɾ /ʹlida/ [ʹɫidɐ] boy; ɔυo:υɾ /ʹkubud/ [ʹkubud˺ ~ ʹkubud] supporter.

/ɟ/ This is generally a voiced palatal stop [ɟ] but it shows allophonic as well as speaker variation in much the same way as /c/. Allophonically this becomes [ɟ-] when followed by a non-front vowel, so when followed by /a/ or /u/. Many speakers also slightly affricate this sound to [ɟʝ ~ ɟ-ʝ] in pre-stress position. The pronunciation [ʝ] may also be heard in any position but is generally stigmatised. It may occur initially in unstressed shortened versions of some pronouns and is also possible in loan word which begin with a voiced palatal. In this position voicing starts prior to the stop closure and manifests itself as prenasalisation. Its usual position is intervocalically or in a voiced environment and as such historically was an allophone of /c/ but degemination and shortenings have allowed for its phonemic status. It never occurs in word-final position. Examples: nnń /ɟa/ [^ɲɟ-ɐ] it.visible.3; υωn:ω /ʹlinɟin/ [ʹlinɟin] south-adj; ιnυ /hiɟla/ [hiɟlɐ] empty; ɔʌn /ʹkiɟa/ [ʹk̟iɟ-ɐ] anvil; ɔʌv:ω /ʹkiɟin/ [ʹkiɟin] anvil-adj.

/ɡ/ This is a voiced velar stop [ɡ] which has allophonic variation in much the same way as /k/. This sound becomes [ɡ+] allophonically before the front vowel /i/ and as such it forms a near merger with [ɟ-] mentioned above, though the two sounds occur in different environments and are recoverable if the following vowel is altered. It does not occur in initial position. Its usual position is intervocalically or in a voiced environment and as such historically was an allophone of /k/ but degemination has allowed for its phonemic status. It never occurs in word final position. Examples: vɔɷ /ʹjugma/ [ʹjugmɐ] beech; ɔυɔ /ʹkuga/ [ʹkugɐ] wall; ɔυɔ:ω /ʹkugin/ [ʹkug+in] wall-adj.

Nasals

/m/ This is the voiced bilabial nasal [m]. It becomes slightly devoiced when preceded or followed by /h/, but it is not fully devoiced. It may occur in both onset and coda position. Examples: ɷɷɷ /ʹmuma/ [ʹmumɐ] mouth; vɷι /ʹjumha/ [ʹjumhɐ ~ ʹjum̥hɐ] bread; vιɷ́ /juʹham/ [juʹham] egg.3.

/n/ This is the voiced alveolar nasal [n]. It becomes slightly devoiced when preceded or followed by /h/, but it is not fully devoiced. It may occur in both onset and coda position. In the standard language, unlike in many other languages, it does not assimilate for place of articulation when followed by non-alveolar consonants. Examples: ωωω /ʹnina/ [ʹninɐ] smell; ɔιω /ʹkuhna/ [ʹkuhnɐ ~ ʹkuhn̥ɐ] to jump; ɷωo /ʹminba/ [ʹminbɐ] tale; ɷɾώ /miʹdan/ [miʹdan] tongue.3.

Fricatives

/s/ This is the voiceless alveolar fricative [s]. There is some speaker variation as to its exact production. Women generally produce this sound apically with some retraction [s̺] whereas men generally produce this laminally, [s]. It may occur in onset or coda position. Examples: ccc /ʹsisa/ [ʹsisɐ ~ s̺is̺a] liquid; ɷcn /ʹmisca/ [ʹmisc̠a ~ ʹmis̺ca] spit; ɔoυ:c /ʹkublis/ [ʹkubɫis ~ ʹkubɫis̺] again.

/h/ This is the voiceless glottal fricative [h]. It remains voiceless in all positions, including intervocalically. It may occur in onset and coda position. Examples: ιιι /ʹhuha/ [ʹhuhɐ] air; vɷί /juʹmah/ [juʹmah] bread.3.

/z/ This is the voiced alveolar fricative [z]. There is some speaker variation as to its exact production. Women generally produce this sound apically with some retraction [z̺] whereas men generally produce this laminally, [z]. It may only occur in voiced positions, so intervocalically and between a vowel and a voiced sound and as such may usually be treated as an allophone of /s/, but degemination and loss of coda /r/ and /l/ followed by /s/, which is allophonically [z], have altered its status slightly, though it is less phonemic than the voiced stops as it has not resulted from pronoun shortenings. It does not occur in final position. It is shown in Romanisations as 'z'. Example: ɔcʌ /ʹkizra/ [k̟izɾɐ ~ k̟iz̺ɾɐ] cliff.

Liquids and Approximants

/r/ This is a voiced alveolar trill [r]. In initial pre-stressed position and intervocalically it usually has at least two taps in all other positions it is more commonly a voiced flap [ɾ]. In the environment of an adjacent /h/ it will be slightly devoiced, though not fully so. Examples: ʌʌʌ /ʹrira/ [ʹrirɐ] noise; oʌ́ω /ʹparna/ [ʹpaɾnɐ] bad.2; υιʌ /ʹluhra/ [ʹɫuhɾɐ ~ ɫuhɾ̥ɐ] father; υιʌ́ /luʹhar/ [ɫuʹhaɾ] father.3.

This trilled sound is notoriously difficult for many lula to produce and often becomes other sounds, which would usually be classed as speech defects. One very common ‘speech defect’ is to produce this sound as a ‘whistled s’, i.e. an apical retracted [s] which sounds a little like a kettle boiling. Another common pronunciation is to produce the trill as an apical voiced fricative [z̺]. The flapped [ɾ] allophone does not undergo these changes.

/l/ This is a voiced velarised alveolar lateral [ɫ]. It may appear in both onset and coda positions. In the environment of an adjacent /h/ it will be slightly devoiced, though not fully so. Examples: υυυ /ʹlula/ [ʹɫuɫɐ] person; υύι /ʹlalha/ [ʹɫaɫhɐ ~ ɫaɫ̥hɐ] alone.2; υɔύ /luʹgal/ [ɫuʹgaɫ] house.3.

/j/ This is a voiced palatal approximant [j]. In coda position it is produced with slightly more friction [j ̝]. It may appear in both onset and coda positions. In the environment of an adjacent /h/ it will be slightly devoiced, though not fully so. Examples: vvv /ʹjija/ [ʹjijɐ] animal; v́ιɷ /ʹajhum/ [ʹaj ̝hum ~ ʹaj̊ ̝hum] egg.1; ɔιv /ʹkuhja/ [ʹkuhjɐ ~ ʹkuhj̊ɐ] leg; υɔυ:v /lugʹlij/ [ɫugʹɫij ̝] house-lat.

Vowels

The language can be either analysed as having two phonemic vowels, one high and one low, or three, where the high vowel is split into two separate phonemes. The vowels are predictable based on the consonant which follows, with [i] appearing before coronals, and [u] appearing before non-coronals, and [ɐ] being produced if there is no following consonant. These three realisations may be considered as a single vowel, /V/, although we shall see that the status of [ɐ] is usually considered separately to the other two. There is also a 'phonemic' low vowel /a/ which occurs as a grammatical marker in the object forms of sentences which shows agreement with the subject. The phonemic status of this vowel is also open to debate. In the Romanisation of the language the three vowels 'i', 'u' and 'a' are given, with [ɐ] being shown as 'a'. Note that [ɐ] only occurs in unstressed position and [a] only in stressed position, but because of the proximity they are often considered as a stressed and unstressed allophonic pair.

Taking this analysis, the three vowels are described below.

/i/ This is a high front unrounded vowel [i]. It is consistently high and front in both stressed and unstressed positions. Examples: ɔɾɔ /ʹkiga/ [ʹk̟igɐ] axe; ɔɾɔ:ω /ʹkigin/ [ʹk̟ig+in] axe-adj.

/u/ This is a high back (slightly) rounded vowel [u]. It is consistently high and back in both stressed and unstressed positions, though it may variably be produced with less rounding tending towards [ɯ] before non-labial consonants. Examples: ɔυɔ /ʹkuga/ [ʹkugɐ (~ ʹkɯgɐ)] wall; ɔυɔ:ɷ /ʹkugum/ [ʹkugum (~ ʹkɯgum)] wall-ben.

/a/ In stressed position this is a low mid vowel [a]. In unstressed position this is more centralised to [ɐ]. Example: oίn /ʹpahca/ [ʹpahc̠ɐ] happy.2.

In linguistic discussions the phonemic status of the vowels is much discussed, especially the status of [ɐ]. In a 'neutral' word, or the root pronunciation of the word a following coronal consonant will mean that a vowel is produced as [i]. (Note here that /l/ does not behave like the other coronals and so its velar gesture is considered of greater importance than its coronal gesture.) If a following consonant is non-coronal then it is produced as [u], with slight or no rounding. If however no following consonant is present we have the production of [ɐ]. In an analysis such as this [i], [u] and [ɐ] are treated together. [a] is only ever produced as a marker of person on the object to show agreement with the subject. As every sentence necessarily has agreement with the subject, this vowel is present frequent, but it serves a grammatical role as opposed to a phonemic one. If it is considered non-phonemic, then the phonology can best be described as having just one vowel.

Historical changes have meant that the vowels have become phonemic. A word such as ɔʌι work with the root consonants K-R-H is pronounced [ʹkihɐ]. The vowel [i] is pronounced because of the following /r/ which has disappeared because it is in coda position. This has therefore led to [i] being followed by the non-coronal [h]. This could on different levels be analysed as /kVrhV/ from a purist perspective. Phonological processes would analyse the first /V/ as [i] and the second /V/ as [ɐ] whilst the /r/ would disappear in coda position. Such an analysis is preferred by many as it best demonstrated the underlying consonants, such that the related word ɔʌ́ι [ʹkaɾha] may be analysed as /karhV/. There are however certain isolated irregularities in the production of the vowels, such that the word onυ may be pronounced as expected [piɟlɐ] and here it means 'book', however it has the irregular pronunciation [puɟlɐ] meaning 'ghost'. Because of such changes, it may now be considered phonemic. In such a case [a] and [ɐ] are usually considered allophones of the same vowel.

Prosody

Stress

Stress is usually placed on the first syllable of the root of a word. Exceptions include where a grammatical /a/ is inserted. This will always take the stress. It usually occurs in the first syllable of the root anyway, but in agreement with the third person it will occur on the second syllable of the root. Suffixes generally do not affect stress placement, though the lative suffix :v /-ij ~ -ja/ used inflectionally will always be stressed, and the negative suffix :ɔ /-uk ~ -ka ~ -ga/ may be stressed if the negative itself is stressed.

Intonation

Main sentence stress falls on the part of the sentence where agreement with the subject falls. This however is not the case if this word is a pronoun (which in non-formal speech are shortened). If there is a pronoun then the stress will fall on the verb, and if there is no verb, then the stress will fall on the subject. There is a rise in intonation before the main sentence stress and then the intonation falls from this main stress. There is no difference between statements and questions.

Phonotactics

The phonotactics of a word are very simple. Mila syllables may be of the form (C)V(C). Syllabification starts from the root of a word and base forms will be CVC.CV. If the grammatical /a/ is added then this will be factored into the syllabification such that we have aC.CVC, CaC.CV, CV.CaC. Prefixes are always of the form VC. If this appears directly before a consonant in a root then we have the form VC.CVC.CV, for example. If this is before the vowel /a/ or another suffix then the consonant will resyllabify as the onset of the following syllable, giving for example V.CaC.CVC. Suffixes are treated as a continuation of the syllabification process in forming the root. When two consonants are across a syllable boundary they will assimilate for voicing, such that a voiceless obstruent may not appear next to a voiced consonant (with the exception of /h/). Otherwise any combination of two consonants is possible.

Orthography

The orthography in Mila is an Abjad system, that is each letter represents a consonant and vowels aren’t shown, the vowels are however predictable from the consonants but a few rules are necessary. It was historically accurate when created, but later changes in the language mean that reading needs to be learnt and there is not always a one-to-one equivalence. Shortened pronouns are written in their long forms even when produced in their shortened form and irregular verbs are written as if they are regular, but need to be learnt. These are, however, the exceptions and the majority of cases can be predicted from the orthography.

The letters (in alphabetical order) are as follows with their corresponding phonemic equivalent:

c	ɔ	υ	v	ɾ	n	o	ɷ	ω	ʌ	ι
/s (~ z)/	/k (~ g)/	/l/	/j/	/t (~ d)/	/c (~ ɟ)/	/p (~ b)/	/m/	/n/	/r/	/h/

One thing which is immediately obvious from the table is that the voiced and voiceless obstruents have just one symbol for both forms. The environment of the letters in most cases, though, will tell us whether the sound is voiced or voiceless.

In determining the pronunciation of a word syllable structure needs to be adhered to. All roots of Mila are triconsonantal, thus the word for good is oιc [puhsɐ] where the letters represent the roots P, H, and S. To form the correct pronunciation we start from the left to determine the first syllable. Onsets are more important than codas and so the first letter o /p/ will form the onset. Onsets can be maximally one consonant long, so we then will need a vowel (which at this stage will be represented by ‘V’). We now have pV. We can now add the next consonant, ι /h/ to this syllable to close it giving us the first syllable pVh. This syllable is now complete and nothing can be added to it as it has maximally one onset consonant and one coda consonant.

We now need to start to form the second syllable with the next letter c /s/ which becomes the onset. As a syllable necessarily has a vowel this is then added to form the second syllable /sV/. As no more consonants are available we then have the completed word /pVh.sV/, where the ‘.’ shows the syllable boundary. As was stated above the main stress is on the first syllable of the root and so will fall on the first syllable here.

We now need to determine the value of the vowels. This is predictable based on the coda consonants of their syllable . If the coda is [+lab] or [+vel] (cf. the distinctive feature table in 1.2) then this vowel will be /u/. If the coda is any other consonant it will be /i/. If however there is no coda present this will be /a/.

In the first syllable of our example we have /h/ as the coda consonant so the vowel will be /u/. In the second syllable there is no coda consonant so this will be /a/. This gives us the full pronunciation for oιc as /ʹpuh.sa/.

In an example such as ncn meaning the colour silver we can again start from the left to form the syllable outlines /cVs.cV/. The first syllable ends in /s/ so the vowel in this syllable will be /i/ and once again the vowel in the second syllable is not followed by a consonant so will be /a/. So ncn is pronounced [ʹcis.cɐ]

There are, however, a number of phonological rules which need to be added to this basic outline.

Voicing

The first is that adjacent consonants must agree in the amount of voicing. If therefore a voiceless obstruent (apart from ι /h/) is adjacent to a voiced consonant then it will become voiced. The word for house is υɔυ before voicing considerations are taken we have the basic outline of /ʹluk.la/ (the asterisk indicates a wrong form). The /k/ needs to agree for voicing with the following /l/ which is voiced. The /l/ cannot devoice so the /k/ must become the voiced /g/. giving us [ʹlug.lɐ]. Similarly, the word for more is oωɾ which gives the basic outline /ʹpin.ta/ but the /t/ needs to agree for voicing with the preceding /n/ so will become /d/, giving us [ʹpin.dɐ].

Liquid Elision

If coda /l/ or /r/ occur after /u/ or /i/ respectively then the /l/ and /r/ will be elided. We thus have the example of the word language ɷʌυ which has the root MRL. Without elision the syllabification gives /ʹmir.la/, but the /r/ follows /a/ and is in coda position, so the pronunciation is [ʹmi.lɐ]. Similarly the word for four is oυɾ which will syllabify prior to elision as /ʹpul.da/ but with elision it surfaces as [ʹpu.dɐ].

Degemination

If two identical consonants occur across a syllable boundary within the same word then this will degeminate, i.e. be pronounced as a single consonant. Note that voicing assimilation will occur before degemination, such that same is coo syllabifies as /ʹsup.pa/ prior to degemination but surfaces as [ʹsu.pa] but the /p/ will not subsequently voice.

Note that a combination of this rule with liquid elision means that the pronunciation of some words have merged in their base form. We have for example to sleep is ιυι which syllabifies as /ʹhul.ha/ but the liquid elision rule means it surfaces as [ʹhu.hɐ]. The word for air is ιιι which initially syllabifies as /ʹhuh.ha/ but then degeminates to [ʹhu.hɐ] being pronounced identically with the word for to sleep.

Grammatical /a/

A grammatical /a/ can be inserted (which will take the word stress). This is shown in the orthography by placing an acute accent above the consonant it precedes. I am hungry is for example ɷ́on. As the acute accent is above the ɷ we must insert /a/ before this sound. This gives us the modified root of aMPC. Starting from this /a/ we then syllabify /ʹam.bic/. You are hungry translates as ɷón will insert /a/ before the second syllable giving the modified root of MaPC which syllabifies as /ʹmap.ca/. And finally (s)he is hungry translates as ɷoń giving the modified root as MPaC. Again starting from the left we start with /mV/ The vowel will be /u/ because of the following non-coronal sound, but the /p/ will not complete the syllable, but rather forms the onset of the following consonant, giving the syllabification of /muʹbac/.

Irregularities=

There are a number of words with irregular pronunciations which are not indicated in the orthography and a learner must just learn them. The word onυ can mean both book and ghost. For the meaning book it is pronounced regularly as [ʹpiɟ.lɐ]. For the meaning of ghost it is pronounced irregularly as [ʹpuɟ.lɐ]. It is because of such words that /i/ and /u/ are generally accepted as separate phonemes, however the amount of such irregularities are very limited.

Verbs have often become irregular in their past or future tenses. This is also not shown in the orthography, so the verb to sleep ιυι has the present tense form as the regularly pronounced ʌ:ιυι [iʹhu.hɐ], but the irregularly pronounced past and future forms (spelt regularly) as ω:ιυι [inʹhu.hu] and ι:ιυι [uʹhu] respectively.

Romanisation

The Romanisation of Mila follows more closely the final pronunciation than the underlying phonological structure. So degeminated consonants as well as elided liquids are not shown, whilst voiced obstruents are shown as voiced. Unstressed [ɐ] is shown as 'a'. Stress is not normally indicated, as it is usually predictable based on the word root, but it may be indicated with an acute accent above the stressed syllable. The romanisation has the advantage that it allows for irregularities to be shown, such that the past and future tenses of to sleep as given in the previous section are shown as 'inhuhu' and 'uhu' respectively.

Morphophonology

Mila has a series of affixes, either prefixes or suffixes. In the orthography these are shown separated from the root with the symbol :. Syllabification of suffixes is a continuation of the syllabification process of the root. Prefixes are syllabified in the form VC. A word such as closed-adj is ʌ:cιɾ:ω which syllabifies prior to phonological processes as /irʹsuh.tin/ but will surface as [iʹzuh.tin]. Because of the syllabification rules any affix may have a number of allomorphs. The dubitative suffix is :o /-p/ which may surfaces as [-up] if following an onset consonant, [-pa] if following a voiceless coda and [-ba] if following a voiced coda.

Morphology

Word formation in Mila is based around the semantically relevant triconsonantal roots. Words rarely occur in their base form and will usually be inflected to show their part of speech. As series of affixes are attached to show case.

Nouns

The following is an overview of the affixes which can be used to modify nouns.

Subject

The subject of a sentence is indicated with the suffix :n [-ic ~ -ca ~ -ɟa]. The subject in Mila usually indicates the actor of a sentence, although when an intransitive verb is used which modifies another verb the subject of the sentence may appear in the predicate position. As such the language may be analysed as showing syntactic ergativity.

υɾn:n-ɾɾɾ́-ʌ:ɷωʌ

[ʹlit.cic da iʹmin.ra]

child-sub it.present.3 pres-cry

This child is crying.

Locative

The locative is used to indicate position, both physical and temporal and can be translated as at, in, on, by, near etc. It is formed by the prefix ʌ: /i(r)-/ placed directly on the noun it refers to. It never takes stress and so the stress will remain on the syllable it would appear on if the prefix were not present.

onυ:n-ʌ:ɔυɷ-nnń

[ʹpiɟ.lic iʹgu.ma ɟa]

book-sub loc-table it.visible.3

The book is there on the table.

Supraessive

The supraessive prefix ɔ: is used to indicate position above the noun it modifies. As with the locative it does not affect stress placement. So ɔ:ιnc /ukʹhicsa/ 'spre-cloud' means simply above the clouds. It usually refers to position without physical contact, though occasionally it may have a similar meaning to on top of or at the top of. An example of this is ʌ:ɔυn /iʹguɟa/ which, by using the locative here, means at/on the mountain. By using the supraessive, though, we have ɔ:ɔυn /uʹkuɟa/ which would mean at the top of the mountain. Another example would be ʌ:ɔυɷ /iʹguma/ which can mean on the table but could equally indicate at or by the table using the locative, but ɔ:ɔυɷ /uʹkuma/ takes away the ambiguity and can only mean on (top of) the table.

It can also be used for figurative purposes, so a king could be spoken about as being ‘on top of the people’ ɔ:υυυ /ugʹlula/. Some of these words have taken on derivational meanings, so the example just given ɔ:υυυ also means haughty and as such has become confused with the identical verbal prefix ɔ: (see below) which is used for iteration, such that the homonym ɔ:υυυ can also mean crowded.