Bäladiri/Reconstruction

Bäladiri is a Westerlander language of the central lowlands. It is part of the lowlands Sprachbund. It features prominent consonant gradation and vowel shortening rules. It has a complex phoneme inventory, with contrasting vowel and consonant length. Its phonotactics are fairly simple: syllables with simple onsets and codas are the most abundant, though sequences like <-ghðr-> do appear as a result of vowel shortening. The direct ancestor of Bäladiri is what I'm going to call pre-Bäladiri, an agglutinative language with fixed word-inital stress. That stage of the language is from before entry into the Sprachbund.

Bäladiri is marked most prominently by its gradation system. Diachronically speaking, it is old enough to have been analogised out in many places (sometimes giving forms like agháinda as opposed to *aáinda/*âinda, of the verb  "see sth.") but young enough to 1) still be acutely productive 2) not cause significant fusion, thus preserving an agglutinative morphology 3) feature multiple grades with fairly straightforward marking.

Bäladiri lenition is caused by many factors. The two most prominent ones are nasal suffixation and stress shifts, respectively softening the consonant they lenite by one and two grades. Nasal suffix lenition is the younger of the two, having been generalised within vague living memory of elderly speakers.

The last stage before the formation of modern Bäladiri can be called pre-Bäladiri (pBd); it indicates the stage of the language as it was before it underwent the characteristic stress shift that caused it to undergo lenition. This stage of the language is, thus, from before it started participating in the Lowlands Sprachbund (LSB).

Five developmental stages can be identified between pre-Bäladiri and Bäladiri before nasal lenition. They can be grouped into two classes — Lowlands and Westerlands (WSB) — based on the Sprachraum of origin. They are:
 * 1) stress shift & pitch accent development (LSB)
 * 2) posttonic lenition (LSB)
 * 3) baritonic vowel shortening (LSB)
 * 4) intertonic vowel shortening (WSB)
 * 5) phonotactic metathesis (WSB)

The progress of the steps is indicative of Bäladiri's uncertain and waning position in the LSB even after its initial integration.

Considering these changes, working back we get the following consonant and vowel inventory:


 * /pp p b tt t d kk k g ʔ/ 
 * /f v θ ð s z ʂ ʐ x γ/ 
 * /mm m ɳɳ ɳ ɲɲ ɲ ŋŋ ŋ/ 
 * /rr r ll l jj j/ 
 * /ts tts dz ddz tʂ ttʂ dʐ ddʐ/ 


 * /i i: y y: u u:/ 
 * /ɛ ɛ: ø ø: ɜ ɜ: ʌ ʌ:/ 
 * /a a:/ 
 * /ai au ɛi/ 

This phonemic inventory of pBd had a frequency distribution that was weighed more strongly in favour of "harder" obstruents and with far more long vowels. The vowels represented by  were lower than  in the vowel space, as shown by their differing reflexes after the third and fourth stages, and those same reflexes indicate that  patterned with  rather than the back vowels, possibly being more front than in Bäladiri (u > ï; o > ä, a > ë).

Pre-Bäladiri stress was fixed on the initial syllable of the root, barring prefixes. This stress wasn't characterised by any one feature at first, but acquired a high pitch value immediately before the first developmental stage. This anticipatory stage, though phonemically irrelevant, had big phonetic and diachronic consequences. Graphically represented it took the form of:
 * > mạvas --> mávas

This anticipatory stage was probably carried out under the influence of an Adaric tongue or language cluster, and it functioned as a gateway towards stress protraction towards the end of the word.

The first stage was carried out using a few guidelines:
 * 1.1. all stress shifts leave low tone on the initial syllable and all syllables inbetween
 * 1.2. stress in roots is shifted to the penultimate syllable which acquires high tone
 * 1.3. all syllables beginning in "hard" consonants draw stress
 * 1.4. final syllables beginning in "soft" consonants (/w j γ/) do not draw any stress and are instead disregarded in syllable counting
 * 1.5. medial syllables beginning in "soft" consonants often draw stress and are disregarded only occasionally and irregularly

This stage essentially produced some form of a pitch accent system in Bäladiri where accent was still predictable and was based on one downshift (where the ancestral word had a high tone) and an upshift (on the antepenultimate or the ultimate). It made changes such as:
 * > mávas, mávatīs --> mávas, màvátīs
 * > máθul, máθultīs --> máθul, màθúltīs
 * > ákāl, ákājas, ákājlī --> ákal, ákājas, àkā́jlī

The second stage followed imminently; there are very few written records that demonstrate stage one without stage two, leading many to believe that they were probably concurrent and happened in parallel. It was carried out relatively simply:
 * 2.1. consonants directly following a baritonic vowel are lenited by two grades
 * 2.2. consonants directly following an acrotonic vowel are not lenited
 * 2.3. consonants that start a non-initial prefix or clitic when a prefix chain is present, or are the onset of a root, are lenited by one grade

Whereas the first two substeps acted as a one-time change, the third substep is still an active process in modern Bäladiri, giving rise to formations such as:
 * > avárkës <-- *a-bárkës ( <-- *abárkas)

This stage further modified all words that experienced such stress shifts, giving stem alternations characteristic to Bäladiri the beginning of their shape. It made changes such as:
 * > mávas, màvátīs --> mávas, màátīs
 * > máθul, màθúltīs --> máθul, màúltīs
 * > ákāl, ákājas, àkā́jlī --> ákāl, ákājas, àghā́jlī

This shift accounted for forming two stems per word in a predictable manner.

The third stage followed the second stage and probably started out as a way to resolve the amount of hiatuses that formed as a result of lenition-triggered deletion and spread out from there. Although deceptively named baritonic, its scope was marginally wider. It operated under the following principles:
 * 3.1. the first baritonic vowel (i.e. the downshifting vowel) acquires a peaking contour and becomes acrokinetic
 * 3.2. the acrotonic vowel (i.e. the upshifting vowel) acquires a dipping contour and becomes barikinetic
 * 3.3. all baritonic vowels get shortened by one degree
 * 3.3.1. all short baritones get reduced
 * 3.3.2. all long baritones become short
 * 3.4. the barikinetic vowel gets shortened by one degree in the same way as baritones do (§3.3)
 * 3.5. the acrokinetic vowel does not get shortened
 * 3.6. ultimate vowels in disyllabic words get shortened in the same way as baritones do (§3.3)

This formation of two contours and a low-high difference prompted another pitch system reorganisation that led into the fourth stage that worked like so:
 * 4.1. all vowels between the acrokinetic and barikinetic get shortened by one degree
 * 4.1.2. all reduced baritones get clipped
 * 4.1.2. all short baritones get reduced
 * 4.2. the barikinetic vowel loses its accent and becomes generically low
 * 4.3. the acrokinetic vowel loses its contour and becomes an acrotone/upshift

These two steps working in tandem gave a chain shift that produced results on a gradient resembling:
 * > ákāl, ákājas, àghā́jlī --> ákal, ákajës, aghájli
 * > aghā̀ðáito, aghā̀ràðálaī --> aghðáitä, aghrðáläi
 * > màátīs, màúltīs --> màë́tis, màḯltis
 * > ákāūn --> ákaun

Combined with lenition-conditioned deletion, this step featured a clean-up in the form of diphthong and sequence contraction. Some of these contractions took the shape of:
 * au > ō
 * awa > ā (ákār < *ȧkāwār)
 * {aë aï} aä > au ā (máutis < *mȧvatīs)
 * jä jë jï > i (-ājas > -ais)
 * ji wu > i u
 * wä wë wï > u (-āwas > -aus)
 * aj > i (not in the first syllable of the root)
 * {äi ëi} ïi > e i (aghrále < *ȧkārālajī)

All resulting diphthongs are considered to have an offglide and, if stressed, carry stress on their first syllable.