Bäladiri/Reconstruction

Bäladiri is a Westerlander language of the central lowlands, part of the northern branch of Assaric languages. 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 category, 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-initial 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 iu > ō ø̄
 * awa iwa uwa > ā jā/ē vā/ō (ákār < *ȧkāwār)
 * {aë aï} aä > au ā (máutis < *mȧvatīs)
 * jä jë jï > i (-ājas > -ais)
 * ji wu > i u
 * ëwë > a (-tale < *-ta-walajī)
 * wä wë wï > u (-āwas > -aus)
 * aj > i (not in the first syllable of the root)
 * {äi [ëi} ïi] > e i (aghrále < *ȧkārālajī)
 * uä uë uï > ū (sûnas < *sùë́nas < *su̇ŋanās ←— *su̇ŋas)

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

The fifth stage followed shortly after the loss of medial vowels and the creation of complex clusters. It is separated from the other stages — instead of being conflated into the fourth like vowel coalescence — because it gave marginally different results across Bäladiri dialects.

The core coastal dialect clustered all the fricatives together and rearranged clusters so as to give an ascending gradient of sonority. This phenomenon was most prominent in imminent momentane verbs; their 2, 3 and 2 forms feature clipping that produces the following metatheses:
 * *ȧkārātawalajī > *àkā̀rā̀táwalajī > *àghā̀rā̀ðáwalaī > *aghrðále > aghðrále
 * *ȧkārātawasājī > *àkā̀rā̀táwasajī > *àghā̀rā̀ðáwasaī > *aghrðáse > aghðráse
 * *ȧkārātawajlī > *àkā̀rā̀tàwájlī > *àghā̀rā̀ðàájlī > *àghrðëíli > *aghrðíli > aghðríli

Some Bäladiri dialects coalesce the fricative and trill of this verb into a fricative, somewhere between [ʒ] and [ʐ] (giving semi-opaque verb forms such as ažðále), while others drop the first (velar) element. Core Bäladiri most frequently employs metathesis, even in situations where it would be impractical.

Many Bäladiri dialects are undergoing more lenitions: the core and closest peripheral Bäladiri dialects feature a nasal-conditioned lenition process, where the nasal element causes a single lenition step in some consonants (obstruents, geminates) and not others (sonorants and voiced fricatives that were already weakened) — this phenomenon is still underway, but its current partial implementation forms an isogloss centered around the core of Bäladiri.

Pre-Bäladiri morphology was far more regular and agglutinative: almost all segments were still fully analysable, with clear-cut patterns and concatenative tendencies. Its direct ancestor was proto-North-Westerlander.

Pre-Bäladiri nominal morphology was a partial devolution of its ancestor's: already in pBd the characteristic Westerlander noun classification system was in the process of falling apart. Pre-Bäladiri nouns were characterised, as Bäladiri nouns still are, by a stem and its characteristic nominative suffix. Whereas the class suffix was a feature of a noun's entire paradigm in proto-North-Westerlander, in pre-Bäladiri it was already limited primarily to the nominative; nouns that were derived from the same root but had a different class suffix were identically declined outside the nominative. A small amount of Westerlander class suffixes survived as derivational morphemes, with examples such as pBd <*-džu>, a diminutive, and <*-ann>, originally a familial/affectionate marker that became a part of quite a few opaque formations.

A standard, regular Pre-Bäladiri noun with its segmentation: >> *suŋas —— brother  << -         SG           PL  NOM   suŋa-s     suŋa-t-īs ACC  suŋa-n-ās  suŋa-t-ās INS  suŋa-n-ā   suŋa-t-ā VOC  suŋa-a-ī   suŋa-t-a     [ —→ súnge, sûtë] OBL  suŋa-Ø     suŋa-t-Ø

The ergative case of its descendant, Bäladiri, was a full instrumental in pre-Bäladiri.

The contamination of the nominative plural with an epenthetic <*-ī-> is a remnant of a generalised plural of a specific set of noun classes. The apparent irregularity of the vocative and oblique singular (no -n- singular demi-morpheme) has its origins in a quirk of Westerlander morphology: proto-North-Westerlander used to mark definiteness with a nasal suffix which was, whilst still productive, dropped out of the vocative as it was implied by the case. The vocative, over time, acquired an intensifier <*-ī> (unrelated to the epenthetic vowel) that ended up grammaticalised onto it. The oblique, with origin in an adverbial case, never took the definite suffix as-is. The presence of this nasal demi-morpheme in the nominative suffix is mostly a feature of the case's characteristic suffix.

Some peripheral Bäladiri dialects have regularised the paradigm by reanalysing the nominative, vocative and oblique singulars as having a nasal: this gave the proto-forms ,  and <suŋan>, which then gave peripheral Bäladiri <sungënis>, <sungëni> and <sungën>

This new singular marker ended up patterning with the plural <*-t-> and, defunct as a category in pre-Bäladiri but still present in a few fixed phrases, the dual marker <*-p->, both of which were pre-existing and firmly established elements of the language's morphology.

A standard, regular Pre-Bäladiri verb with its segmentation:

>>>         *akādējī —— to see         <<< --- --> INF SG           DU            PL  NOM  ak-ā-dej-ī      ////      ak-ā-dej-t-īs     ["see sth."] ACC ak-ā-dej-n-ās   ////      ak-ā-dej-t-ās INS ak-ā-dej-n-ā    ////      ak-ā-dej-t-ā VOC ak-ā-dej-ī      ////      ak-ā-dej-t-a OBL ak-ā-dej-Ø      ////      ak-ā-dej-t ....................... --> PRES MOMN 1ST  ak-ā-ūn      ak-ā-wār      ak-ā-jas 2ND  ak-ā-l       ak-ā-laī      ak-ā-jlī 3RD  ak-ā-m       ak-ā-saī      ak-ā-was ....................... --> PRES CONT 1ST ak-ā-tā-ūn   ak-ā-tā-wār   ak-ā-tā-jas 2ND ak-ā-tā-l    ak-ā-tā-laī   ak-ā-tā-jlī 3RD ak-ā-tā-m    ak-ā-tā-saī   ak-ā-tā-was --- --> IMMI MOMN 1ST ak-ā-rā-ūn   ak-ā-rā-wār   ak-ā-rā-jas 2ND ak-ā-rā-l    ak-ā-rā-laī   ak-ā-rā-jli 3RD ak-ā-rā-m    ak-ā-rā-saī   ak-ā-rā-was ....................... --> IMMI CONT 1ST ak-ā-rā-tā-ūn ak-ā-rā-tā-wār ak-ā-rā-tā-jas 2ND ak-ā-rā-tā-l ak-ā-rā-tā-laī ak-ā-rā-tā-jlī 3RD ak-ā-rā-tā-m ak-ā-rā-tā-saī ak-ā-rā-tā-was --- --> PAST MOMN 1ST ak-ā-ja     ak-ā-jato      ak-ā-jal 2ND  ak-o†       ak-ā-to        ak-ā-l 3RD ak-ā-jan    ak-ā-janto     ak-ā-wann ....................... --> PAST CONT 1ST ak-ā-tā-ja  ak-ā-tā-jatä  ak-ā-tā-jal 2ND ak-ā-t-o†    ak-ā-tā-to    ak-ā-tā-l 3RD ak-ā-tā-jan ak-ā-tā-janto ak-ā-tā-wann

Pre-Bäladiri regularly derives from the same proto-North-Westerlander stock that Bäladiri's siblings, such as Bassani and Burri, do too. This ancestor of pre-Bäladiri can for ease of reference merely be called proto-North-Westerlander (pNW).

The latest change, still in progress in late pre-Bäladiri, was the shuffling of the back vowel system. Pre-Bäladiri had in its vowel system a vertical arrangement of back vowels /*u *ʌ *ɑ/, which was the result of a reorganisation of very late proto-North-Westerlander back vowel system of /*u *o *ʌ *ɑ/. This system ended up reorganising fairly late, after all the (relatively) earlier vowel shifts affecting late pNW. The last stage of the shifts can be described in a series of steps:


 * 1.1. Proto-NW  dissimilates to /*a/ if another /*ʌ/ is already present in the word (usually in a more stressed position). This is very indicative of the lack of rounding in both /*ʌ/ and /*a/.
 * 1.2. Proto-NW /*ʌ/ rounds to /*o/ after certain consonants, generally alveolars. It is probable that they had contrastive rounding; remarkably, this never happened to the vowels before rounded consnants, probably indicating the rounding had elements of an offglide.
 * 1.3. Proto-NW /*ʌ/ rounds to /*o/ next to labials, similar to what happened to it after rounded consonants.
 * 1.4. Proto-NW /*o/ shifts unconditionally to /*u/, merging with it.
 * 1.5. Proto-NW /*u *ʌ *a/ shift cleanly to pre-Bäladiri <u o a>

By cross-comparison with other North-Westerlander languages, we can see that the pre-Bäladiri word <suŋas> stems from a very late proto-NW root that had the shape of <*sʷʌŋʌ->. It is very likely that the alveolars unrounded in pre-Bäladiri, though it is impossible to say exactly how monolithic this process was.

Some complications arise with the long counterparts of those vowels, which were rarer in proto-NW than in pre-Bäladiri, which abounds in them. The developments surrounding these are, in rough order:


 * 2.1. The long proto-NW vowel /*o:/ shifts to /*u:/ unconditionally relatively early on.
 * 2.2. Proto-NW /*ʌ:/ shifts to /*o:/ after rounded consonants, in parallel with §1.2. This doesn't extend to rounding next to labials other than /*w/, and even then only when preceded by it. This new /*o:/ very rarely shifts to /*u:/ (in certain irregulars).
 * 2.3. Proto-NW /*ʌ:/ shifts to /*a:/ in all other environments unconditionally.
 * 2.4. Proto-NW /*o:/ unrounds to /*ʌ:/, becoming the long counterpart to /*ʌ/.

Certain environments are known to interfere with the generality of these shifts, though the exact context is hard to determine. The shift of /*ʌ/ to /*o/ next to labials is generally blocked whenever the labial is surrounded by two /*ʌ/ on either side. This is very evident in the pre-Bäladiri word <gopor>, which comes from proto-NW <*ⁿkʌpʌʀ>. It is also known that velars (such as the <-*ʀ> in the aforementioned word) occasionally blocked the dissimilation of /*ʌ/ to /*a/.

Comparison between the North-Westerlander languages shows that proto-NW had a consonant system much unlike Bäladiri. A subset of the most likely ancestral system would be: - ######## |  Labial  #     Alveolar    #  Velar  # - - - - -|- - - - - # - - - - - - - - # - - - - -   Nasal  |    m     #   n        nʷ   #    ŋ    # Tenuis P | p   pˀ  # t   tˀ   tʷ  tʷˀ#  k   kˀ # Voiced P |   b     #   d        dʷ   #    g    # Prnas. P |  ⁿp     #  ⁿt       ⁿtʷ   #   ⁿk    # Fricatv. |   f     #   s        sʷ   #    x    # Approx. |   w     #   ð        ðʷ   #    ɣ    # -