Previous Up Next

1.2  Diachronic view

1.2.1  Consonants and general root structure

Talmit preserves the ancient Proto-Tallic phonotactics almost completely intact. If one abbreviates

then the possible Proto-Tallic root types are (all monosyllabic):

The only phonotactic restriction is that Si and Sf/Ss have to be different, thus no **√grar or **√glal. The plosives can be identical, and often are, as e.g. √paps ’bake’, √tatn ’take, seize’, √kekr ’wine’.
Very rarely, m is observed instead of an initial plosive (√me ’human being’), s as an initial sonorant (√ksatr ’fly’, √psar ’rub’), w, j as final sonorants (√dew ’advice’). The breaking of the otherwise strict pattern suggests an ancient loaning.

Talmit preserves the initial clusters almost without change. The most apparent one is that aspirated stops become spirants: *ph, *th, *kh > ϕ, θ, χ.
Also notable is the unusual shift *pw, *bw, *phw > *pw̃, *bw̃, *ϕ > pn, bn, ϕn where the w was pronounced nasalized in order to dissimilate the homorganic compound. It probably first happened in *mw > *mw̃ > mn, resulting in a u̯|n alternation as táuma ’shield’ and támne ’protection’ (*tawma, *tamwə < √tamw). Compare also the alternation -wa|-ma ’-hued’ (the former after vowels, the latter after consonants), both regular reflexes of *-ŋwa (in Kymna -ngva).
In dialects, one observes *phw > [f] or [p͡f] (an affricate as German pf); *bw > [v] or [b͡v].
In addition, tl, dl have naturally merged into affricates, and khl, khr, khw, khj yielded voiceless hr, hl, hw, hj respectively. Palatalized dentals changed to affricates: *tj > θ and *dj > *[ʣ] > z:

plainptkbdgϕθχ
+ rprtrkrbrdrgrϕrθrhr
+ lpltlklbldlglϕlθlhl
+ wpntwkwbndwgwϕnθwhw
+ jpjθkjbjzgjϕjθjhj
+ spssks

In combination with the secondary sonorants n, s, the clusters *tn, *dn became *[ʦn], *[ʣn] > zn; and ts became ss – the standard dialect shows a dislike for the affricate [ʦ]. But of course, where θ is realized as [ʦ], θn merges with the result of *tn.
The fricative [s] is naturally voiced in contact with voiced stops. Note also that x just represents /ks/ as the Romans intended it to.

plainptkbdgϕθχ
+ npnznknbnzngnϕnθnχn
+ spsssxbzdzgzϕsθsχs

Final plosives + secondary sonorants did not change, except for *mw > mn. Instead of *zs one always finds ss and is doubtful whether something as awkward as *zs actually existed:

plainmz
+ rmrzr
+ lmlzl
+ wmnzw
+ jmjzj
+ smzss
+ nmnzn

The phoneme *Q is, similarly to the Indo-European laryngeals, one of unknown realization which disappeared in all Tallic languages. It was in any case a velar, uvular or a pharyngeal sound – perhaps a voiced velar fricative [γ], a uvual trill [ʀ] or a glottal stop [ʔ] – but more probably various sounds in allophonic variation which is now beyond recovery. In Talmit, *Q becomes a copy of the following voiceless sound (similar to the Japanese bound moraic phoneme spelled with a subscript tsu), changes to g before nasals and z; to k or χ before other sonorants, and becomes -t finally (probably via *-k):

The development probably went *Qn, *Qm > *ŋn, *ŋm > gn, gm, as Talmit dislikes the velar nasal (similar to the Slavic languages except Polish) and eliminates it whenever possible. For the same reason, *ŋχ, *ŋχw, *ŋχr, *ŋχl generally experience metathesis to χn, χm, χn, χn; only ŋk, ŋg are stable in Talmit.

If the word already contains k, *Q dissimilates to p|b. If a labial is appended, it dissimilates to t|d, hence krat ’mountain’ (< *kraQ), krápjo ’few mountains, a couple of mountains’, krádmi ’many mountains’.

1.2.2  Sonorants in contact

Talmit avoids clusters of sonorants and converts them into geminates. They assimilate according to the sonortiy hierarchy

so that *nr > rr, *nl, *rl > ll.

On the other hand, s simply becomes voiced and leads to nz/zn, rz/zr, lz/zl, zm/mz which are permitted. In combination with m one finds that rm/mr, lm/ml and mn are permitted; and *nm usually becomes mm. If more nasals are present, *nm changes to rm or lm instead: *dilon-men > dilórmen ’long/deep sleep’ (cf. Latin germen < *gen-mṇ, carmen < *can-mṇ). In a number of cases, *nm can also experience metathesis to mn.

Where clusters like ln appear in the language, they are simplified from *lmn and the like.

There is evidence that Proto-Tallic additionally had a second rhotic phoneme *R as a rare sonorant (perhaps the flap [ɺ]). It usually became r in Talmit, but dissimilated to l in presence of another r, and to n in presence of both r and l.
In very ancient times, this sound might have been just a splitting consonant inserted between two vowels, as it only appears before suffixes and after prefixes. For example, the conculsive verbal ending -un becomes -nun after vowels, as in gánun < √ga, which may be explained by *ga-R-un with assimilation to the following nasal (contrast: gárun < √gar). The causative prefix has the form sar-/sal-/san- which might be the result of *saR before vowels, later generalized to all positions by analogy.


The development of vowels is best discussed in light of word derivation. The Proto-Tallic roots could take several grades which are presented below.


1.2.3  E-grade

The e-grade was originally formed by appending the schwa ə to the root. For verbal roots, it conveyed the notion of an organized way of doing an action, or of an associated abstract noun.

In Talmit, the schwa changed and affected the preceding vowel in a process which can be termed evening-out and is close to the Germanic and Welsh a-umlaut. Namely, preceding high vowels i, u were lowered to e, o accompanied by ə > e. But unlike a-umlaut, preceding e, o could also be lowered to a accompanied by ə > a. Preceding a, on the other hand, could be raised to e (with ə > e), but quite often it was the schwa which simply changed to a. Thus the vowels were evened out in height, the resulting combinations being o-e, e-e, a-a, and rarely o-o:

This process was usually resisted by u when near labials (pu, ub, um etc.), and by a when near velars (ga, ka, aχ etc.):

From here, and from the very common -e after o, e, the ending -e was usually substituted by analogy for a-a in other cases as well, hence e.g. táple instead of *tápla.
In a few cases, the various results of the e-grade were subject to selection by different contexts. For example, from the root √tatn ’take, seize’ one has the back-formation tázne which has retained the meaning ’taking, seizure’ while the umlauted form tézne has acquired the meaning ’reading, counting’ and is by analogy treated as if it was from a root *√tezn, hence tézne ’reading, counting’, ténza ’an event of reading or counting’.

Further examples of e-grade nouns:

1.2.4  Zero grade

The zero grade was formed by the bare root alone. It could only be formed from ”heavy” roots with a final sonorant, the sonorant thereby becoming syllabic. It denoted the instrument, tool or appartus needed to perform an action (in particular the senses, for which see 3.4.2). Talmit always merges agents with instruments:

As it can be seen, syllabic sonorants broke into combinations ’vowel + sonorant’, where the inserted vowel tended to be of the same height, but was very often a before r, and i before l. There is considerable dialectal variation here: tébin, gwímur, tápal and héker can also be found, for example.

An invention of Talmit was the extension of the zero grade to ”light” roots. They appended a vowel + -s by analogy to ”heavy” roots with s:

Note that there are three different words for ’mouth’ in Talmit: bno is used for the physical organ only; tápil ’mouth, sayer’ and aχágas ’mouth, eater’ can be used in this sense as well, but in different contexts – tápil might be used in the sense ’shut your mouth = stop talking’, aχágas in the sense ’shut your mouth = stop champing’. Unlike bno, however, they also denote agents, as in ’truth-sayer’ and ’man-eater’. Compare English ’foul mouth’, ’a mouth to feed’.

Monoplosive verbal roots do not have a zero grade, but they can be easily compounded, e.g. √thol ’serve’: θólda ’servant’ (-da ’intelligent being’), √gi ’see’: gipnós ’sense of sight’ (pnos ’sense, emotion’).

1.2.5  Fortified grade

The fortified grade is sparsingly used for ”light” biplosive roots with a sense similar to the zero grade, but with an augmentative shade of meaning. The root-final consonant is geminated, -e < *-ə is suffixed (no word can end in a plosive) causing evening-out (1.2.3), and b, d, g, z are are devoiced in the gemination to pp, tt, kk, ss.

1.2.6  We-grade

The noun *wə simply meant ’place’, but was grammaticalized in Proto-Tallic. On the one hand, it became the nominative marker for the durative aspect (its reflex is T. wa, the nominative marker for verbal predicates 2.2.3), on the other hand it was agglutinated to nouns as a locative case.
However, the two often collided, e.g. √ka ’earth, ground’: *ka-wə=wə ’ground-on=nom’ (i.e. ’the one on the ground [did somethig]’). In these cases, the agglutinated *wə instead jumped as an infix into the root itself in the inverted form *əw1. This had iconic significance – how to better indicate that an object is inside something than by an infix? Hence: *kəwa-wə.
This process ceased to be productive in Talmit and was lexicalized only in a couple of words.

In ”heavy” roots, *wə remained a suffix, but was appended to a metathesized root form (itself hailing from ancient and not recoverable sound changes), e.g. √podj ’flowing water’: *poj(ə)d-wə ’river-in’ > T. páidwa ’gone, vanished, dissolved state’, as in the saying ples-páidwa lit. ’a drop in the river’ = ’gone without trace, vanished from the face of the earth’.

Some common we-grade words are:

1.2.7  A-grade

Proto-Tallic has always been a ”nominophile” language, preferring nouns over verbs (e.g. ’happiness’ instead of ’be happy’; a simplex meaning ’sitting’ instead of ’to sit’ etc.). But no judgement – even if you are verbophile yourself, probably some of your best friends are nominophiles, right? In any case, at an even earlier stage the language must have been like Japanese or Korean with regard to nouns – they had no plural and were basically all mass nouns. The distinction between tangible or discrete objects (like ’cloud’) and abstract notions (like ’cover’) was very much smeared out.
However, Proto-Tallic started to form a singulative from these nouns (cf. Welsh adar ’birds’, sing. aderyn ’bird’) simply by agglutinating the numeral *aQ ’one’ in the reverse form *Qa1. It assumed the same infixed position as the we-grade in the case of monoplosive roots. Intervocalic *Q became weakened to *h and ultimately disappeared, so that the typical development was:

The resulting contact aa was altered to [æa], spelled ea:

Thus arose an apparent a-infixion pattern (see 1.1.3).

For ”heavy” biplosive roots, *Qa was appended to the metathesized variant, later simplified to -a:

The primitive a-grades must have had an inserted schwa to confirm to the open-syllable rule of Proto-Tallic, hence the syllable division *ka-wə-tQa. The schwa disappeared in Talmit and Kymna, but was retained in another Tallic language Hadam where it generally became i, as evidenced by H. kabit ’tree’ < kawət-, pl. kavtam (root √katw).

Note also that *Q was regarded as a sonorant, and in fact as a consonant corresponding to a, in symmetry to j, w:

syllabicnon-syllabicconsonantal
ij
uw
aQ

The a-grade was also applied to verbal roots where it denoted a count noun associated with the action. This was often just ’the act of doing’ or a decomposed part of the action, but could also be the object of a transitive verb, a thing produced by the action, and so on:

Finally, for biplosive ”light” roots, *Qa was also simply appended, but it diphthongized with the preceding vowel:

The same diphthongization happened to j and w in medial position as well (cf. the Greek i-metathesis *pheresi > φέρεις ’you carry’ and the Welsh plurals):

This might account for the origin of the metathesized roots.

1.2.8  I- and u-infixed forms

The Proto-Tallic language had corresponding i- and u- dipthongs for all five vowels: *ai, *oi, *ei, *əi, *ui and *au, *ou, *eu, *iu, *əu where i and u became ə to avoid the awkward ii̯, uu̯. They were mostly used to mark evidentiality on verbs, but came to denote tense in Talmit. It largely preserves this system, but changes:

These changes are marked by the very same evening-out rule as before (1.2.3), except for ai, au, oi which are permitted. Again, some dialects are more radical here and only allow diphthongs of the same height, changing the others: *ai > [æ], *au > [ɔ], *oi > [œ].


Previous Up Next