Evolving verbs

[Jonlang]

At the moment I have two conlangs I'm actively working on: P and L because I still haven' named them. Anyway, I know how I want the verb systems to work in each, I'm just unsure how both systems could evolve from a proto-language, so I'm looking for help. I'll begin by giving some information on both conlang's verbs:

P

The verbs of P are heavily based on Welsh (literary Welsh, mostly). Every verb has a stem to which suffix can be added to mark person, number, tense, and mood, etc. So, for example, the stem dwel- (run) can yield dwelan (I run/I am running), dwelamp (we run), dwêl (he/she/it runs) [suffix lost, marked by long vowel on the bare root], dwelag (they run), dwelant (verbal noun - functions as an infinitive, gerund, and present/active participle). This stem can also take subjunctive suffixes which are, in fact, just the subjunctive forms of 'to be' suffixed to the verb stem: dwelsain (that I run), etc. There are also imperatives and a passive/past participle which are added, as well as indicative forms for preterite, imperfect, perfect, pluperfect (the present is actually non-past and is used for the future, too), but they're all suffixed to the stem. The stem is usually the part of the verb left when the verbal-noun suffix is removed, i.e. dwelant > dwel-, but not always, sometimes the verbal-noun form is also the stem, e.g. ethech (play) > ethech-

L

The situation in L is much different because... well... when I began work on L it was kinda the "anti-P". It was to be another conlang where I could use other sounds and sound combinations that I love the sound of that didn't fit P: e.g. P has lots of voiced plosives and fricatives, so L has very few voiced plosives and far fewer fricatives (with different distributions to P), but has a lot more voiceless plosives, a lot more Ls, and more "vowely" words. Where P is VSO in normal word order, L is SOV in the literary register and SVO in the spoken register, but the word order is freer than in P.

Anyway, the L verbs aren't as complex as the P verbs. They only agree in number, so I, you (sing.), he/she/it all use the same form of the verb, as do we, you (pl), they. L also has dual number which is lost in P. So the verbal stem in L lúelant- (run) becomes lúelanta (present sing.), lúelantoa (present pl.), etc. and L doesn't mark the S or the O on the verb, unlike P which marks the S, i.e. one cannot tell from lúelanta who is running.

The idea is that L is probably more conservative (generally, as a language) than P, as L preserves the proto-language's subjunctive forms and imperative forms (replaced in P by forms derived from the irregular 'to be').

Of the two, I am more willing to make changes to L as P is my "main" conlang. My usual system is to create what I want in each conlang and see how I could justify these in the proto-language, but I'm finding it quite difficult with these two verbal systems. Obviously, I'll answer any questions if it helps - I've had to type this in two sittings so I'm unsure if it's clear.

[Ares Land]

How about this?

L is the more conservative of the two; so really person marking is a later innovation -- stemming for earlier cliticized pronouns / auxiliaries.

Here's a very rough idea of how it could work:


The basic forms are:
Singular: duelant
Plural: duelantoa

Now, let's say
a) nt > n word finally.
b) u > w before a vowel
c) while we're at it: oa > a

Sg: dwelan
Pl: dwelanta

Now let's add some personal pronouns:
1sg. a
1pl: pa
3pl: ka

Word order is still SOV but pronouns are backed.
dwelan a 'I run'
dwelanta pa 'we run'
ki dwelan 'the dog runs'
dwelanta ka 'they run'

Word-final vowels are lost; pronouns cliticize
nt > n before a consonant; then nasals assimilate.
dwelana
dwelampa
ki dwelan
dwelaŋka > dwelaŋa

n is lost word-finally; final vowels are lost ; ŋ > g

dwelan
dwelamp
ki dwela
dwelag

Final vowels are once again lost, with compensatory lenghtening:

dwelan
dwelamp
ki dwêl
dwelag