A well formed Tanaþekńabmi syllable has either CV or CVC structure. In affixes, non-morphological heads (morphological head: the rightmost derivational morpheme or the root if there is none), and borrowed words a C nucleus is permitted. Note: derivational morphemes differ from inflectional morphemes in that derivational morphemes change the category of a word. In no circumstances can a syllable be without an onset. Diphthongs, multi-segment onsets or codas, and vowel hiatuses are not tolerated.
[top]Allophony, Syllabification, Vowel Harmony, and Infixation
The allophony is best explained in the terms of the constraints posited by Optimality Theory as opposed to the limited rule-based SPE. Allophony only occurs in whatever is not the morphological head in a word. Vowel harmony occurs everywhere except the primary stress and affects the roundedness feature of metrical feet. Infixes are malformed prefixes moved by phonological rules. Relevant generalisations are listed below:
Always have onsets.
Onsets must be consonants other than glides.
Onsets and codas must only have one segment.
Diphthongs are banned.
Adjacent vowels are banned. If they occur, then epenthesise the least marked consonant [m].
Metathesis is banned.
Fusion is banned.
Vowel-breaking/unpacking is banned.
The voice feature of an onset cannot change.
The place feature of a segment in a morphological head cannot change; epenthesise [+low] which is [a] when a coda is in danger of assimilation to a labial/labio-dental.
Stressed vowels in the input must retain the same roundedness feature in the output.
The right-most segment of a word (unless it is a reduplicant) cannot change. If keeping the anchor violates the ban of complex onsets/codas, epenthesise [+low] which is [a] between the violating features.
If two adjacent segments not in a morphological head have differing place, the latter segment assimilates to the to the place of the former if possible (dentals and alveolars are treated as having same place). Labials and labio-dentals do not assimilate and instead epenthesise [+low] which is [a] between them.
Elide all high vowels from non-morphological heads, except if elision causes two identical consonants to be adjacent.
After all processes, pick the most sonorant nucleus according to the sonority hierarchy (from least preferred to most preferred):
*NUC/stop ≫ *NUC/fric ≫ *NUC/nasal ≫ *NUC/liquid ≫ *NUC/glide ≫ *NUC/VHI ≫ *NUC/VLO
Prefixes must align to the left edge of the stem (stem: root+suffixes). If malformed, then infix to the leftmost location in the stem that allows repair.
After picking the most sonorant nuclei, assign onsets to them.
After assigning onsets, assign codas.
If there is a stranded onset, then epenthesise [+low] which is [a] as the nucleus.
Tanaþekńabmi is a stress-timed language with trochaic feet (stress-unstressed), a near exhaustive parsing (not every syllable has to be in a foot), and left-to-right directionality (creating feet from left to right). The primary stress falls on the leftmost foot of the morphological head of a word and all other stress is secondary. Relevant generalisations are listed below:
Feet must be binary; all feet must consist of two syllables.
Parse every possible syllable into a foot that is binary.
Feet must be aligned to the left edge of a word.
All segments in a foot must be round if one segment is round.
The primary foot must be aligned to the left edge of a morphological head.
The stressed syllable of a foot must be aligned to the left edge of the foot.
Constraints are used in Optimality Theory to pick the most optimal surface candidate out of a variety of inputs. They can either check for faithfulness to the input or markedness in the output. A constraint dominates (≫) another if a violation in the dominating constraint knocks out a candidate and another candidate that violates the dominated constraint is preferred. The ranking of constraints cannot be known (, ) in some cases. Accounting for richness of the base, theoretically any input or loanword can be transformed into a well formed Tanaþekńabmi word.
ONSET: assign one violation for each syllable without an onset.
MHd-IDENT[Place]: assign one violation when a segment in the input of a morphological head has a different place feature in the output.
IDENT[labial]/[labio-dental]: assign one violation for each segment that was labial/labio-dental in the input but changed place feature in the output.
ONSET-IDENT[voice]: assign one violation for each onset that differs in voicing from input to output.
HNuc: assign one violation for each syllable nucleus that is less sonorant than its other parts.
MHd-MAX: assign one violation for each segment in the morphological head that is there in the input but missing in the output.
IO-ANCHOR-R-σ: assign one violation for each rightmost segment in a non-reduplicant word that is changed or missing.
LINEARITY: assign one violation when the order of some segments change.
UNIFORMITY: assign one violation when two or more segments are realised as one.
INTEGRITY: assign one violation when a segment is realised as multiple segments.
*ONS/glide: assign one violation when an onset is a glide; by sonority hierarchy also when a liquid or vowel.
*VV: assign one violation when two vowels are adjacent.
*HETERORGANIC: assign one violation when two adjacent segments have differing place features.
*DIPHTHONG: assign one violation when a diphthong occurs.
*COMPLEX: assign one violation when an onset or coda has more than one segment.
IDENT-σ́-IO(round): assign one violation if a stressed vowel in the input has a differing roundedness feature in the output.
MHd-DEP: assign one violation when epenthesis occurs in a morphological head.
MAX-C: assign one violation when a consonant in the input is missing in the output.
OCP: assign one violation for every pair of adjacent identical segments.
*NUC/x: assign one violation when a nucleus is x manner feature.
DEP: assign one violation when epenthesis occurs.
IDENT[Place]: assign one violation when a segment in the input has a different place feature in the output.
MAX-V: assign one violation when a vowel in the input is missing in the output.
ALIGN-L(Prefix, Stem): assign one violation for every segment from the left edge of the stem (stem: root+suffixes) the closest edge of a prefix is away from. This allows infixing to repair malformed prefixes.
NOCODA: assign one violation for every syllable with a coda.
CONTIG: assign one violation if two segments are adjacent in the input but not in the output.
Here is an example tableau showing the syllabification and allophony of the word Tanaþekńabmi /tana-θe-kini-bini/. All affixes in this word are not derivational. (Imagine dotted lines separating columns that have unknown ranking)