I remember I had read something about a disappearing *-i in pre-Etruscan before. I started thinking about Miguel Carrasquer Vidal, a man with a fierce sense of logic with which I had many an online arguement with beforehand. Regardless of our differences of opinion, I respect him as a keen thinker in the realm of comparative linguistics and someone who typically backs up his views with a wealth of solid references. In this post on sci.lang in October 2004, Vidal had mentioned this interesting tidbit which he based on his readings of Robert S. P. Beekes, the Emeritus Professor of Comparative Indo-European Linguistics at the University of Leiden who has also written about the origin of the Etruscans (see pdf) and Pre-Greek substrate (see pdf). Miguel stated:
That doesn't change anything in the analysis of -le < *-la-i, or -ls < *-la-si. If it makes you feel better:At the time I had first read it, it didn't "make me feel better", nor did other people completely embrace the idea as you can read by the responses. While it's acceptable to suggest that Etruscan had some sort of prior syncope (before the documented one that occured around 500 BCE), suggesting also that an ablative *-si-si was reduced to -is by way of a convenient one-time case of added haplology seemed to me ad hoc. The package as a whole then came across as just a shade too mathematical, assumptive and logically problematic to accept.
- genitive *-si, *-ia-la > -s, -(ia)l
- ablative *-si-si, *-ia-la-si > -is, -(ia)l(a)s
- dative *-si-i, *-ia-la-i > -si, -(ia)le
Yet now, here I am before my computer, and after hammering out my Etruscan database, I have to admit with tail between my legs that Miguel may not have been so crazy afterall, although his point of view needs clearer rules and further examples. It's not enough to say that vowels disappeared in final position because it clearly wasn't the case in passive participles in -u (e.g. caru "made"), as an example.
It seems to me that we might be able to theorize the following to add to Miguel's ideas above concerning an early stage of pre-Etruscan:
- There were once 3 possible vowels in word-final position: -a, -i and -u.
- Both -a and -i were dropped while -u remained.
- Former a-stems and i-stems however still kept their vowel when suffixes were added.
More food for thought.
 The loss of -a can be seen in the word un 'libation' (< *una) with the locative une (< *una-i) attested in the Liber Linteus (LL 8.xvii, 10.xxxiv). If the vowel had not been there, we'd expect *un-i in the locative instead.