Basque butterflies live!

[ECOLING at aol.com]

I am very happy to see the continued discussion of various
Basque words for 'butterfly'.

I intend to prepare an integrated treatment of where the
evidence and the discussion have led, as far as I understand,
but that will have to wait for late in February.

Here I will merely mention some salient points which bear on
the lines of reasoning, without citing again the Basque forms
which we all have in our archives of these messages.

And to clarify at the outset,

Larry Trask's goal at the moment is something he defines very narrowly
and precisely, as looking for native monomorphemic lexical items
which are the best candidates to have been present in pre-Basque
or ancient Basque (I don't remember which term to use here),
approximately 2000 to 1700 years ago, a time at which
one can distinguish at least many loans from Latin or Romance.
That excludes all verbal roots, for example.
The way Trask implements those goals, his (to me excessively strict)
notions of evidence, not preponderance but restricted to certain
kinds of evidence, lead him to exclude much relevant data.

My goals are (as concerns Basque) to encourage explorations of
its early stages, using careful linguistic
techniques and archaeological etc. techniques as guides.
My goals are (as concerns words for 'butterfly') to use the knowledge
I have of these combined with what Trask has so generously provided
us to provide a "best estimate" of which of those forms might
with high probability go back to earlier Basque forms,
whether those earlier Basque forms were ultimately borrowed or not.

Here are the contributions to the topic that stand out for me:

***

1.  Roz Frank just pointed to Basque scholarship which differs with
Trask's view, in that words with initial /p-/ are not thereby automatically
excluded from being considered as possibly ancient Basque.

***

2.  I have noted in some of the extensive private communications Larry
and I have had on 'butterfly' that Japanese permits expressive words
with initial /p-/ such as /parapara/, while not allowing it in ordinary
vocabulary and I think not in earlier loans from Chinese, for example.

***

3.  Larry points to the tendencies of the L. dialect of Basque to have
forms with initial /pin-/ or /pan-/ plus labial, as indicating that
L. dialect forms for 'butterfly' must be innovated there because
different in that dialect.  Linguistic method does not require that
conclusion.  All we can say, much more conservatively, is that
the ASPECT of the L. words for butterfly which begin with
initial /pin-/ or /pan-/ plus labial may not be old, it could easily
be innovated within L. dialects in accordance with their general
tendencies.  So it is quite legitimate to project backwards to
possible (ONLY possible) earlier forms which lacked one or another
of these characteristics.  The fact that certain aspects COULD
have been innovated within L. dialects does not prove they all were.
Perhaps some partial resemblance of an earlier word for 'butterfly'
to such L. dialect preferred templates led to their attraction all the
way into the core of the template.  So it is perfectly possible
(though not provable) that L. dialect

pinpirin < *pitil...  (taking the rule backwards by removing the /n...n/
                                and by removing the reduplicated /p/;
                                on /l/ and /r/ please see below).

***

4.  Larry pointed out in private communication that ancient Basque
*l became /r/, so he argues forms with /l/ cannot be ancient.
On further questioning,
he noted that some Basque /l/ does derive from some other kind of *l
which might have been geminated or something
(I am not quoting in any of this,
simply repeating from memory, and some of this was in private
communication so Larry would have to state this publicly in whatever
words he finds most precise, I cannot do it for him.)

If that is the case, then we could have the /r/ of the L. dialects
going back to earlier simple /l/.  That does not solve all of our problems,
I mention this only to make it clear that the reasoning is, as so often,
very complex, and conclusions not at all cut-and-dried, not even
with the best linguistic techniques.  All rediscovery of the past shares
such dilemmas.

***

5.  Larry pointed out in a recent public message that
*t > Basque /tx/ is quite normal.
This rather strongly supports the possibility at least that the quite
prominent forms for 'butterfly' with /tx/ might once have had *t.
(Larry did not mention this in our private communications, so far
as I remember.)
So we can consider the following not unreasonable as a
tentative hypothesis for projecting backwards at least one sound change.

pitxilota < *pitilota

***

6.  Larry pointed to /txitxi/ 'chick' (again citing from memory,
pardon if I got the form slightly wrong) as an exceptional item,
standing out by a mile, not merely by being one of the few words
for young of animals not formed with a derivational suffix,
but also with its double voiceless stops, not fitting what Larry
believes is the canonical form for earlier Basque.

If however these supposed "exceptions" are gathered together,
then /txitxi/ and /pitilota/ or /pitxilota/ may both be part of a
group of expressives, which *did* allow exactly that canonical form.
By not attempting to exclude so much, but attempting to find multiple
clusters of similarly structured forms which may have been in earlier
Basque, we at least avoid the risk of excluding that which truly *was*
in earlier Basque.
To be fair to Larry, it must be pointed out that he gives evidence
that earlier Latin etc. loanwords with two voiceless stops in initial
positions underwent changes, did not preserve them when brought
in as loanwords.  If this were shown also generally for loanwords
which were expressives, I would concede that it would be a strong
argument against assuming an ancient Basque *pitilota loanword.
Hence the relevance of the items under 1. and 2. above.

***

7.  Larry recently noted that quite a number of words, perhaps
he was talking mostly about the 'butterfly' words, did not have
identifiable morphemic components, that the suffixes were not
early Basque, etc. etc.  There has I think been some more discussion
on some of these.
    But assuming Larry is correct about this, it then follows
that the 'butterfly' words may be mostly monomorphemic,
so on that criterion he cannot exclude them even from his own
much more limited set of data which his own goals are seeking.
(If monomorphemic, this would be fitting for loanwords
whose internal morphology may not be transparent to speakers of a
borrowing language.).

***

8.  While my estimate is that indeed Greek *ptilota
(Thanks to Steve Long for bringing this to our attention)
is a very good candidate for the origin in an earlier stage of Basque
of modern-day /pitxilota/ and many forms similar to it,
I do not claim evidence for that, only lots of evidence which
would be consistent with that.  Quite a different matter altogether.
Not a hypothesis to be discarded recklessly.

***

9.  There is a great underestimation, among those who want sound
laws to be exceptionless and want evidence always to be completely
conclusive, of the degree of unexplained changes (called sometimes
"irregular", but that usually means we simply don't yet know the rules).

In my first message attempting a partial analysis of the Basque
'butterfly' words, I mentioned the fact that we cannot project
backwards from modern English "sparrowgrass" to its actual
origin "asparagus" because the word has been reformed.
Note that it is a loanword which does not fit English morphology
which has been reformed in this way.

If we allow that among the Basque words for 'butterfly'
many might have been reformed (including some in ways
fully in accord with the preferences such as Larry Trask notes
for L. dialects with /pinp-/ and /panp-/), then we certainly
cannot exclude the possibility of reconstructing earlier Basque
forms as the common origin of quite a range of modern Basque
forms, if we are lucky enough to have a wide enough range
of sources of evidence.  Larry's enormous collection of Basque
words for 'butterfly' gives us the chance of having such a
wide enough range of evidence, and I hope we fully use it.
Words for 'butterfly' in neighboring language families are
also potentially relevant, whether or not the word is borrowed,
and cannot be excluded a priori.

Just so, we cannot by regular sound laws reconstruct
both "asparagus" and "sparrowgrass" back to a common
original form.  But we can reconstruct them back to a common
original form.  In this case we have enough information to do so.

In Basque we lack written records in sufficient quantity,
but linguistic techniques should allow us to make some progress
at least towards plausible hypotheses.

***

10.  As yet another example of just how very different the numerous
descendants of a *single original* may at first sight appear to be,
I cite the various forms of the nursery rhyme which English children use:

For the first half of the first line, the following:

1) Eine meine mine me
2) Eenie meenie minie mo
3) Meeny meeny miney mo
4) One-erzoll, two-erzoll, zickerzoll zan
5) 1-ery, 2-ery, ickery Ann
6) Hana mana mona mike
7) Ene tene mone mei
8) Yan tan tehtera pethera

For the second half of the first line, the following (numbers match),
with translations given where appropriate.
As you can readily see, just as with "sparrowgrass",
there need be no relation of meaning, the reformation is based on sound.

1) Und draust bist du  (and you're out)
2) Catch a rooster by the toe
3) Cache ton poing derierre ton do (hide your thumb behind your back)
4) Bob-tail vinegar little tall tan
5) Fillisin follasy Nicholas John
6) Barcelona bona strike
7) Pastor lone bone strei  (this is German, "pastor")
8) Pimp sethera lethera hovera dovera dick.

The last line contains the Celtic Lake-district sheep-scoring numerals,
      argued by some to be the origin of all of the other versions.
      In Yarmouth, the last three are: "cothra hothra dic".

Given this known example of actual historical descent,
and similar examples on a smaller scale can be multiplied
almost without limit in the annals of historical linguistics,

it seems absurd to me to argue that the much more similar
dialect variations in Basque words for 'butterfly' are absolutely
not derivable from a much smaller number of common origins
(i.e. much smaller than the 9 groupings in which Larry Trask
supplied them to us).

Not that they necessarily *do* go back to a small number of origins
(perhaps in the extreme case seven groups of forms all going back
to one origin, excluding the two obvious items derived from
non-expressive vocabulary).
But simply that this kind of change cannot be excluded absolutely,
based on the kind of reasoning we have had presented to us.

We cannot reject as possibilities of historical descent
the kind of historical descent which we know *does* occur.

So I hold out great hope that we will be able to reconstruct one or
a few ancient Basque words for 'butterfly'.
With a high level of justified confidence in our answer,
though not absolute confidence.

Sincerely,
Lloyd Anderson
Ecological Linguistics