| Proto-Talo-Edastean |
|Period||c. -2500 YP|
|Spoken in||Bwimbai valley|
|Writing system||primitive runes|
|Basic word order||VSO|
Proto-Talo-Edastean is a reconstructed language spoken along the Bwimbai river valley circa -2500 YP, and is the ancestor of the Talo-Edastean languages. Its best known direct daughter was Ndak Ta, which in turn spawned the huge Edastean family of languages.
There is some very minimal attestation of Proto-Talo-Edastean, in the form of a few tantalizing runic inscriptions usually found engraved in stone. Some hundred and forty such inscriptions are known, but all of them are brief, and over a hundred consist of nothing but one or more personal names. Furthermore, the runic system varied by time and place and did not mark vowels, and not one is believed to contain a complete sentence. Thus scholars have gleaned only a precious little information from them, and none of it certain. What follows is instead based on reconstructive work: both internal reconstruction from Ndak Ta and comparison between NT and Tlaliolz.
Sound Changes to Ndak Ta
The sound changes from Proto-Talo-Edastean to the Ndak Ta of Tsinakan's era saw the loss and merger of many dorsal sounds. Some of the changes involved assimilations that behaved in unusual manners, which has been a source of puzzlement for Edastean scholars. There are not a great number of changes, as the time depth between PTE and NT is roughly half a millennium, but least the following ones occurred:
1. *dn, *bm, and *gŋ assimilate to [n:], [m:], and [ŋ:].
2. Unstressed vowels between an intervocalic stop and a final nasal were lost, after which POA assimilation applied and the nasal became syllabic. Example: sikadn "hurricane" < *sikaden
3. Unstressed initial vowels were syncopated, before intervocalic consonants and before nasal+stop clusters. Initial nasals then became syllabic before stops. This appears to account for all the wordforms seen:
mbontai "seed" < *ambóntai (while untai < untai)
baus "ox" < *abáus
oslók "forget" < *oslók (cluster /sl/ blocks syncope)
enánda "feel" < *xenánda (initial consonant blocks syncope)
4. Loss of dorsal fricatives: *x was deleted in all environments. *xʷ lost its fricative component and merged with preexisting *w in all environments. In most cases where a vowel hiatus was created by loss of *x, the second vowel of the sequence was deleted (this rule forbidding most hiatus probably predates proto-Talo-Edastean, and continued to be active in Ndak Ta). Examples: laid "year" < *laixod, and sai "female" < *saxi
5. Vowels in checked syllables (i.e. before a coda) weakened: *a, *e, *i, *ai, and *au were lowered or centralized to [ɐ, ɛ, ɪ, ɐɪ, ɐʊ] respectively. *u and *o appear to have been unaffected. The timing of this rule is constrained by the loss of *x and *ɴ; vowels before coda *x were apparently not weakened, or else a phonemic set of them would have been created when *x was lost - whereas when *ɴ was lost later, the nasalized vowels it left behind continued to display weakening. This rule continued to be active in Ndak Ta.
6. The uvular stops *q and *qʷ merged into their velar counterparts *k and *kʷ. Similarly, clusters of *ɴq and *ɴqʷ merged into their velar counterparts *ŋk and *ŋkʷ. Example: ingkwi "liver" < *iɴqʷi
7. Remaining instances of the uvular nasal *ɴ which had not been clustered with uvular plosives, merged into *ŋ only after *u and *o. In all other environments it was deleted, leaving prior vowels still nasalized and thus creating a set of contrastively nasalized vowels. Once again, in cases where a forbidden vowel hiatus was created, the second vowel of the sequence was deleted. Example: âka "air" < *aɴeka
8. The voiced labiovelar plosive *gʷ was apparently fronted to [bʷ] without losing its labiovelarization. When following a nasal, the entire *ŋgʷ cluster was fronted to [mbʷ]. Example: bwai "star" < *gʷai
The exact nature of this change has long been debated, for it is peculiar in two respects. First, the retention of labiovelarization following such a fronting is nearly unheard of in the known sound changes of any other language. And second, only *gʷ fronted, leaving *kʷ and *ŋʷ alone - more often such fronting involves the entire labiovelar series, though other exceptions are known.
9. Perhaps the most bizarre development in Ndak Ta is the direction reversal of nasal assimilation. In proto-Talo-Edastean, whenever morphology brought together a nasal followed by a stop, the nasal assimilated to the stop's point of articulation, as happens in myriad other languages. At some point in its history, Ndak Ta reversed this, such that stops were assimilated to the prior nasal instead. Note that this affected only those clusters which crossed a morpheme boundary: morpheme-medial clusters had already assimilated in the more orthodox direction, and thus had no other underlying representation on which this rule could act.
It is also debatable whether rule #9 occurred before or after rule #8; while we list it afterwards, it has been proposed that the morphological collision of *m with *gʷ occurred more frequently than with *kʷ, such that the resulting backwards-assimilation of *mgʷ to [mbʷ] spread by analogy to other instances of *gʷ and finally taking the cluster *ŋgʷ with it. If true, this would would reduce the two unusual characteristics of #8 to logical consequences of the other unusual development, #9. It is notable that morphological collision of m with kw in Ndak Ta was sufficiently rare that its outcome (mp? mpʷ?) remains unknown.
10. Voiceless stops became glottal stops before nasals, but only in Kasadgad and the lower Aiwa (during the Akan dynasty of the Ndak empire). Thus some scholars have omitted this from the list of Ndak Ta sound changes, adding it instead to those of early Naidda.
Sound Changes to Tlaliolz
Early Tlaliolz, also sometimes called Common Tlaliolz, was the latest common ancestor of all the later dialects. It was roughly contemporary with Ndak Ta, or perhaps a little later. What is usually called Late Tlaliolz is the array of dialects known to have existed in the classical period. When used without qualification, "Tlaliolz" normally refers to the Eastern variety of Late Tlaliolz, the dialect spoken by the Talo of Huyfárah. It is this dialect which is the best recorded among the lot, just as the Talo were the best-known Tlaliolz subgroup. Words (but no grammar) were also recorded from a western variety, a handful of which found their way into Ndok Aisô, but our knowledge of Eastern is far greater. A handful of Eastern Tlaliolz texts remain on record, and numerous words were borrowed by the Puoni .
to Early (Common) Tlaliolz
1. *x and *xʷ fortified, in initial position only, to [k] and [kʷ].
2. Plosives, both voiced and voiceless, but not *ʦ, became voiceless aspirates after nasal consonants.
3. The diphthongs *ai and *au simplified to [i] and [o]
4. Sonorant consonants (which included *x *xʷ *r *l *w *m *n *ŋ *ŋʷ *ɴ) were deleted in preconsonantal position. A set of long vowels was thus created when vowels lengthened to compensate for the loss of these consonants, and a set of phonemic aspirated plosives was created by the loss of conditioning nasals before plosives. This sound change thus altered the language profoundly.
5. Unstressed initial vowels were lost from many but not all words. This may have depended on the presence or absence of prefixes.
6. *b, *d, *g, and *gʷ devoiced to [p], [t], [k], and [kʷ] in all environments.
After these changes occurred, the phoneme inventory would have looked like this:
|high||i i:||u u:|
|mid||e e:||o o:|
to Late (Eastern) Tlaliolz
Sound Changes to Antagg
The Antagg have received little scholarly attention, perhaps due to their marginal position in Ndak times. However, studies of Antagg texts allow us to tentatively reconstruct the following sound changes between Proto-Talo-Edastean and the Antagg language of c. -1700 YP.
1. Nasals were lost word-finally.
2. Vowel syncope. The exact conditions are not well understood, but it appears to have affected only unstressed monophthongs, and applied in the following environments:
- word-finally after a single consonant
- between */s/ and another consonant.
3. Loss of coda nasals, with compensatory vowel lengthening. This appears to have happened in stressed and final syllables.
4. Stops became unvoiced between a nasal and a stressed vowel.
5. */ɴ/ lenited to /h/, certainly word-initially and probably in all environments.
6. */x xʷ/ fortified to /k kʷ/, certainly after */s/ and probably in all environments.
7. Vowel shift: */a e i o u ai au/ became /a ɛ e ɔ o i u/ in unstressed syllables. Remaining */au/ (and probably also */ai/) became /aː/ in at least some environments.
8. Labialised consonants merged with labials, and remaining uvulars merged with velars.
9. */ts/ lenited to /s/.
10. */l/ was deleted before another consonant.
11. */w/ was deleted after labials, before /u/ and probably before /o/; it merged with */j/ (presumably via */ɥ/) in some environments, apparently before /i e/.
12. */ŋ/ merged with /n/, apparently in all environments.
13. Other changes: /s/ assimilated in voicing to an adjacent obstruent; */r/ became a tap /ɾ/; /ɔ/ may have undergone a shift but remained a distinct phoneme.