One of the problems with my languages is that they don't have much of a 
history. As a consequence, they tend to feel artificial and overly 
regular. I'd like to try to fix that by elaborating the earlier history 
of a few languages and ultimately building language families around each 
one. Tirelat and Jarda are two of the languages I want to start with, 
along with the Sangari roots of Zharranh (a mixed Sangari-Zireen 
language) and whatever source languages may have contributed words to 
Jaghri (an artificial language with Sangari roots).

I know a few things about how languages develop historically, but I've 
been having a hard time trying to work backwards from existing languages 
to reconstruct earlier versions. Still, I think it's worth exploring, so 
here goes.

The consonant inventory of modern Tirelat looks something like this:

p b	t̪ d̪	ʦ ʣ			k ɡ
m	n̪				ŋ
		r̥ r
f v		s z	ʂ ʐ		x ɣ
		ɬ l
w				j

Now what I need to do is figure out how it got that way. I know there've 
been some splits and mergers along the way. One that I've mentioned is a 
phoneme (call it /D/ for now) that merged with /d/ in one dialect and 
/r/ in another.

orig.	dial. A	dial. B
/vidu/	/vidu/	/vidu/
/niDu/	/nidu/	/niru/
/viru/	/viru/	/viru/

One possibility is that this /D/ was actually /t/, and that modern /t/ 
(maybe also /ts/ in some cases) descends from /tʰ/. It's a little 
clearer what might have happened if you look at a broad phonetic 
transcription, where /d/ and /t/ between vowels weakened to [ð] and [d]. 
In one dialect the [d] changed to [ð], and in the other both [d] and [r] 
changed to [ɾ].

orig.	later	dial. A	dial. B
[vidu]	[viðu]	[viðu]	[viðu]
[nitu]	[nidu]	[niðu]	[niɾu]
[viru]	[viru]	[viru]	[viɾu]

So I'll assume for now there was a whole set of voiceless aspirated, 
voiceless unaspirated, and voiced stops in Early Modern Tirelat 
(possibly including a post-alveolar series). It's also likely that there 
was a glottal stop, and corresponding voiced and voiceless fricatives 
for each place of articulation.

pʰ p b	tʰ t d	ṭʰ ṭ ḍ	kʰ k ɡ	ʔ
f v	s z	ʃ ʒ	x ɣ	h

The post-alveolar stops mainly seemed to have developed into affricates, 
either /ts/ /dz/ or /tʃ/ /dʒ/ depending on the dialect.

One sound that sticks out in the modern Tirelat phoneme inventory is the 
voiceless trill /r̥/. This is likely to have come from an originally 
voiced trill /r/, which became voiceless by assimilation when adjacent 
to a voiceless stop or fricative. (This is one reason I've postulated 
the existence of the glottal stop in Early Modern Tirelat.) Indeed, in 
the standard dialect, trills adjacent to voiceless stops or consonants 
are always voiceless, although other dialects allow voiced trills in 
those instances.

It's also possible that there may have been a voiced lateral fricative 
/ɮ/ in EMT, to account for the /l/ ~ /d/ variations between dialects in 
some words, but there may be other ways to account for that.

Approximants /w/ and /j/ are probably reduced forms of vowels 
(especially when you consider that one Tirelat dialect has a phonemic 
/ɥ/, which could be related to /y/). I haven't looked much at the vowels