In Vytautas Ambrazas' Lithuanian Grammar it says "[U]nstressed [long] vowels in Standard Lithuanian show a tendency to be shortened and turn into half-long (sometimes even relatively short) tense vowels. These changes do not harm the phonological system: the contrast is not lost, but only modified. The occasional complete neutralization of the quantity of unstressed vowels can be explained only as a phenomenon of some other (mainly dialectal or sociolectal) phonological system." Some of the examples he gives are acc. sg. duoną vs. nom sg. duona, where the final vowels differ in quantity, and výrų vs. výru, where they differ in both quantity and -- apparently -- quality. What I haven't been able to understand is if the same applies to root vowels, like, e.g., the <e> in kẽlias vs. keliù, or <a> in stãlas vs. stalù (I wonder what happened to the tildes). What say ye, native speakers?

 I won't dare comment such subtlety from a native speaker's point of view, rather I can share my theoretical knowledge.

In standard (as it's taught at universities) Lithuanian only phonological opposition long:short is postulated (and imposed :), all the phonetic (including qualitative) differences being standard secondary subsequencies of this opposition.

In dialects (as 'common' Lithuanian of cities was not explored to such an extent) the following phonetic differences are registered: phonologically long compact vowel (say, /ą/) shows a tendency to get more diffusive; vice versa, phonologically long diffusive vowel (say, /ų/) shows a tendency to get more compact; the exact way differs from dialect to dialect, for example, some Highlands (Aukštaitija) dialects even have something like [u] in the place of /ą/.

As for kẽlias vs. keliù, or stãlas vs. stalù, here we have a different phonological situation (if I got you right): historically short stressed (and therefore prolongated):unstressed (not prolongated) vowels opposition; qualitative phonetic difference is standard: we have respectively [ɛ] and [a] in both cases, the unstressed being somewhat 'closer'.

 

One thing I noticed during the short time I actually had a Lithuanian teacher was that her version of [æː]/<e> in Petras was virtually identical to (my) Swedish allophone of <æ(ː)> before /r/, and from what I've gathered subsequently it probably wasn't a just peculiar idiolectal feature of hers. Now that we have people who know both Lithuanian and Russian on the list it seems like a good opportunity to ask about the realization of <я> between two 'soft' consonants in standard Russian and other varieties in words like пять. How close to standard Lithuanian [æː] would you say that this vowel is (as regards the quality of the syllabic nucleus)? And in what way(s) does it differ?

 As I can work out from introspection, one of phonetic realizations of Lithuanian <e>, [æ] (not [ɛ] that's heard before palatalized consonants) in stressed position can be treated as an equivalent to the allophones of Russian <я> you mentioned, at least within the accuracy of IPA notation. As for differencies, the Lithuanian sound shows a tendency to get more closed; on the other hand, Russian sound is, as usual, enriched with glides and has rather unsharp articualtion - that what, in my opinion, makes the difference. Please note also that we can't talk purely on quality and ignore prosodical moments in regard to Lithuanian.

This is a preliminar answer and implies no warranty. :)

Sergei