In staying intermediate between Celtic and Germanic, Grimm's and Verner's Laws have been conceptualised as a push chain caused by the Proto-Indo-European voiced aspirate series becoming voiced fricatives in most positions (creating an extremely unbalanced system with many voiced fricatives with no voiceless counterparts), a shift shared with Italic (and possibly also Celtic). All affected branches then resolved this lack of balance in different ways.

In Germanic, the voiceless series became fricatives except following another obstruent (these fricatives were voiceless word-initially and after a stressed syllable, but otherwise merging into the voiced fricative series), the plain voiced stop series became voiceless, and the voiced fricatives *β and *ð became voiced stops /b/ & /d/ word-initially (the voiced fricative *ɣ also became a voiced stop /g/ word-initially outside Dutch).

In Celtic, the new voiced fricative series merged into the plain voiced stops. It's usually said that this merged series were stops in all positions in Proto-Celtic, but it has also been suggested that the lenition of voiced stops seen in Insular Celtic already existed on an allophonic level in Proto-Celtic, with the merged voiced series being voiced fricatives in leniting environments, and stops elsewhere. In this view, the development of voiceless fricatives (by lenition in Goidelic, or provection in Brythonic) being caused by the phonemicisation of the previously allophonic voiced fricatives.

In Italic, the voiced fricatives devoiced word-initially allowing the remaining voiced fricatives to be viewed as voiced allophones. Additionally, word-initial *ð (> *θ) merged into *f (comparable to the widespread th-fronting in the Southeast of England). In Sabellic, word-internal *ð also merges into *f, and in Latin most of the word-internal voiced fricatives fortited to stops.

So, in order to remain intermediate between Celtic and Germanic, we have lenited the voiced aspirates to voiced fricatives in most positions (including word-initially). In most instances, voiceless stops develop as in Germanic, but word-initially remain unlenited (as in Celtic). Again, as in Celtic, *p is still lenited, but only to /f/ (whereas there is no clear evidence for any labial feature still being present in the Proto-Celtic reflex of *p outside certain consonant clusters (particularly following *s and preceding a resonant).

Celtic preserves more vowel distinctions than Germanic, and Velgaiɣa is intermediate in that it retains those distinctions in initial stressed syllables, but not in unstressed or non-initial syllables (which are subject to varying levels of vowel reduction, at different stages, as a result of the shift to initial stress). In initial syllables, syllabic resonants broadly behave as in Celtic (albeit subject to the expected vowel reductions when unstressed), but in non-initial syllables, the expected vowel reductions lead to more Germanic-like outcomes. Stressed final syllables have similar outcomes to stressed initial syllables (with a small number of additional mergers/reductions), whilst unstressed final syllables are reduced almost as much as medial syllables, retaining only a few more distinctions.

Many of these historical changes have been obscured by analogy (e.g. the o-stem declension mostly retains the stressed allomorphs, despite being used for all o-stems, even those that etymologically have stressed stems, and so words may have both a stem and ending reflecting the expected stressed reflex).

There is also a regular process of syncope of short vowels in the second of two open syllables when not word-final, but when the vowel that would be expected to be syncopated is a derivational suffix (or part of the formation of an inflectional class of nouns or verbs), the vowel is often restored. This is because that vowel might be expected to lost after light stems, but would be retained after heavy stems, and so the vowel was restored to the light stems by analogy to the heavy stems.

More information on certain words and points can be found throughout the Lexember 2024 posts (although some of these may be a little outdated, as aspects of the language evolved over the course of the month):

1st: rīma "the numerals"

2nd: gānyom "the family"

3rd: gallū "to be able"

4th: klāsmarra "(hill)fort, oppidum"

5th: ɣimmēs "the lunar month containing the winter solstice, ~December"

6th: far-ɣusman "libation"

7th: troggī "nose"

8th: friyos "free"

9th: lustrom "value"

10th: wasū "to dress"

11th: lahtuðaɣa "morality"

12th: an-rīmas "countless, unnumbered"

13th: ɣlohom "glass"

14th: the Universal Declaration of Human Rights

15th: ɣelas "green, blue"

16th: baryai "to die"

17th: skowwunū "to appear as if in a mirror, to show, to reflect"

18th: arguntum "silver"

19th: an-yahtī "illness"

20th: ɣōliryun "vegetable"

21st: ɣardwa "Midwinter, the Winter Solstice"

22nd: ɣostidūriyos "host", ɣostidustrī "hostess"

23rd: bargu "slow"

24th: moiþman "gift"

25th: gentī "birth"

26th: vir-fihtāɣī "hangover"

27th: sudyēs "holiday"

28th: stēr "star"

29th: ambruðai "to rain"

30th: vardūþom "bearded"

There are a few main isoglosses relevant for distinguishing between these varieties

The voiceless labial p:

p: the Eastern varieties generally retain the plosive /p/

f: the Southern and Western Iberian varieties have a fricative pronunciation /f/

fh: Eastern Iberia have /hφ/ for earlier p ~ f

The voiced labial b:

b: Central North Africa, Italy, Gaul, and Illyria retain the plosive pronunciation /b/

v: Iberia, Western Mauretania, and Eastern Libya have a fricative pronunciation /v/

The labial glide w:

w: in Africa, Illyria, and Gaul, the original glide /w/ is retained

v: in Iberia, and Italy the fricative is pronounced /v/, merging with original b in Iberia

Geminates:

CC: in the East, geminates are preserved as geminates

NC: in the Southwest, geminates prenasalise and lose gemination

Emphatic spreading:

C_C̣_C: in the North, emphaticness doesn’t spread

C̣_C̣_C̣: in the South, emphaticness spreads

The guttural h:

h: in the South and East, h is generally preserved

x: in the North, h is pronounced /x/

Ø: in the West h is lost

The hushing fricative š:

š: in the Southwest, the pronunciation of š as /ʃ/ is preserved

s: in the Northeast, š merges into s

Voicing of emphatics:

ṭ: in the South, emphaticness is retained

t: in the East emphatic consonants merge into voiceless consonants

It has a large consonant inventory with labial, dental, alveolar, palatal, velar, and uvular series as well as a couple of glottals. The stops and fricatives come in voiced, voiceless, and emphatic (ejective) series; the alveolar and palatal series have an additional set of affricates (again voiced, voiceless, and emphatic); and the alveolar series also has a set of three lateral fricatives. The labial, velar, and uvular series also have phonemic gaps (all three lack ejective fricatives, and the labials lacks all stops except the voiced stop). There are fairly significant sandhi effects.

The vowel inventory is much smaller. There are 7 different vowel qualities each of which can be short or long, as well as two diphthongs. Syllable structure is CV(C) with long vowels only allowed in open syllables (or if followed by a geminate).

Aud̄d̄aun̄yo has ten different noun classes forming an animacy hierarchy placing the D̄un̄un̄aul (the speaker population, originally conceived as desert-dwelling nomadic human-sized carrion-beetles but now probably no longer beetles to allow for human-like speech) first, with other sapient people, then animate creatures coming before the seven other classes. Mixed groups agree as the most animate member of the group.

There is a certain amount of root+pattern derivation (the main remnant of the Semiticity) from (mostly) triliteral roots for both verbs and nouns, but the system is no longer fully productive with derivational affixes more common in recent strata of the lexicon.

The most important aspect of the morphology is the screeve (a term stolen from Georgian and used rather inaccurately) which defines a scope of relevance for whatever form it is attached to. It is a circumfix marking mood (one of the indicative, subjunctive, jussive, interrogative, imperative, tentative, abilitative, epistemic, or obligative) as well as two tenses (one of ancient-past, middle-past, recent-past, immediate-future, middle-future, distant-future) one for the start of the scope of relevance and one for the end of the scope.

There is an extensive system of fifteen cases, mostly marked with suffixes, but a few with additional prefixes. The dative is used for the object of singly-transitive verbs, the remaining argument of ditransitive verbs takes the thematic case. Additional cases are the genitive, benefactive, aversive, causal, instrumental, privative, locative, ablative, lative, essive, exessive, and translative. Like several other polysynthetic languages, it exhibits suffixaufnahme (i.e. “he shot a man of the people” > “he.NOM shoot.PAST man.DAT people.GEN.DAT”). Nouns may take one or more screeves before or after cases suffixes (or even after some and before others).

The verb complex is much more, well, complex. Ever verb must receive at least one screeve (with two screeves describing “perfect” tenses, three describing “doubly-perfect” tenses e.g. “I have had run” etc etc but more than two screeves being increasingly rare) together with any number of more optional affixes. A verb may take an affirmative or negative prefix, a telicity prefix (the telic prefix is only possible if the two tenses of the screeve are different, and the atelic prefix is only possible if both tenses are different). There are also optional evidential suffixes (distinguishing between direct, secondary, inferential, and hearsay).

Personal pronouns only occur as clitics, which may be used as an enclitic on nouns to mark possession, as an enclitic on prepositions to mark their object, as a proclitic to mark the subject of a verb, or as enclitics to mark the object of verbs. These occur in three persons and singular and plural. The third person clitics occur in all ten classes, the second person occurs in D̄un̄au, sapient, and animate classes and the first person occurs only in D̄un̄au and sapient classes.