Plosive-liquid clusters in euskara borrowed from IE?

[Nicholas Widdows]

<Roz Frank wrote:>

[snip...] sometimes heard
as <bur(u)ko> and at others as <b(u)ruko>. The speaker is fully aware that
the form is <buruko> but that is not what s/he is actually "saying." In the
case of <andere> it strikes me that the same thing happens with the result
being <and(e)re> or <and(e)r(e)a> when the definite article is attached.

</Roz Frank>

The speaker knows (perhaps only unconsciously) that the phoneme sequence is
/buruko/, even if the phone sequence may be [bruko] or [burko]. But (except
in scribe-influenced languages like English or French) the modern reflex is
always a descendant of actual spoken forms. If historical linguists can
state that Old Basque didn't have plosive + liquid clusters then it's
because such occasional phone clusters were never psychologically elevated
to phoneme clusters, and no change of the co-occurrence rule can be
reconstructed from the historical data.

Although, as you say, there are microdifferences in pronounciation that
distinguish native from non-native on what should be almost identical
sounds, these differences are probably held at a broader phonological level.
For example, in my own speech <train> and <terrain> may differ only in that
the [tr] of <train> is an affricate (cf. <czy> ~ <trzy>), at least in quick
speech. Or although I normally say both <dune> and <June> as [dZu:n],
they're not precisely identical, because <dune> feels the influence of its
phonemic form /dju:n/ and can fluctuate in degree of affrication. The
psychological difference is still large.

I would suggest (and I freely admit I have no idea whether this is true, but
I suggest) that to be carried along diachronically a sound reduction of the
kind <and(e)re> or <bur(u)ko> or <t(er)rain> would have to go through a
reinterpretation of its abstract phonemic form. In Old Basque *<andre> broke
a firm rule and would not survive. In modern Basque, does [burko] have the
tap of /buruko/ or the roll of */burko/? In my English initial [tr] can come
from both /tr/ and /t at r/. Once they lose the possibility of reversion to the
old phonology, the vowel can be considered gone.

The non-standard pronunciations can be very divergent. For <I'm going to> I
catch myself saying all sorts of things, [aINg at nt@] and [aNgan@] and
[QNg at n@] and [aNn@], none of which I would want to dignify with a phonemic
form. When they settle down to any one or more forms as stable as <I am> and
<I'm> then they can be entered in my idiolect dictionary.

That said, forms like <gotcha> and <dunno> and <nope> have acquired a stable
written form and can then be enunciated without being the actual slurred
forms that gave rise to them. So this is not to deny that saltations like
<chandra> could have occurred; I just suspect that the ordinary lability of
speech is not ordinarily enough to generate them.

Nicholas Widdows