'butterfly' in Basque, etymologies?

ECOLING at aol.com ECOLING at aol.com
Wed Dec 22 13:29:11 UTC 1999


[ Moderator's note:
  There are a number of new subscribers to the Indo-European list who may be
  confused by continued discussions of Basque etymologies here.  I allow them
  because they often bring up points with respect to methodology that might
  not occur so clearly in Indo-European studies.
  --rma ]

Larry Trask has provided a beautiful collection of words for 'butterfly'
in various Basque dialects.  I received his permission to repost the
contents to the Sound Symbolism list, which has recently collected
words for 'butterfly' from all over the world,
Those interested in sound symbolism or "expressives" are rarely so fortunate
as to receive such a detailed accounting from anywhere.

There is so much information in this data that I believe we can make
a great deal of sense out of it historically, that we can in essence
point backwards towards the forms of words for 'butterfly'
in early Basque or pre-Basque.

At the same time,
I believe these Basque forms show just how fuzzy the borderlines
are between "expressives" and the rest of the vocabulary.

Trask has contributed extensive annotations,
of the kind only a Basque specialist can.
He also may have additional information
on sound changes in the history of Basque which can inform
the considerations adduced here, making some specific hypotheses
more or less likely.

I add below other data and reasoning which are the special focus
of a specialist in expressives, or of a specialist in the kind
of data variation and incomplete information we have here.

I use heuristics based on known kinds of language change
to posit hypothetical reconstructions
of early Basque words for 'butterfly',
to estimate which of the attested forms
are closer to an ancient Basque form,
which have been more heavily altered,
and how each has been altered,
in a few cases also why.

This is not the same thing as following the operation
of regular sound changes, because much of language
change is also analogical change, the mutual influence
or contamination of elements of the vocabulary.
As Malkiel and others have said,
each word has its own history.

Quite independently of the first draft of this message,
without either of us yet having seen the other's notes,
in reaction to Larry Trask's posting of this data set,
Steve Long sent a private message pointing out Greek forms,
including <ptilo->.  I consider this a confirmation of the
results of reconstruction from Basque dialects, and
mention it again at the end of this message with more detail.

I give first  Larry Trask's general conclusions.
Most of those general conclusions seem sensible to me,
except the first one, that the origins of these words must
necessarily be "expressive".  There have been influences
from the kinds of sound patterns found in expressives, certainly,
but that is not the same as attributing the ultimate origins
of the words to such sound patterns.

I obviously accept Trask's accounts of what are common sounds
and syllables in expressives in Basque -- that is his specialty, not mine.
Here is what Trask says, omitting mention of the two items which he
says appear to represent metaphorical senses of ordinary lexical items,
and separating each point he makes to begin on its own line:

************************************************************

"... all the others show unmistakable evidence of expressive origins:

length (four or more syllables);
opaque elements;
frequent presence of the segments <tx> and <m>
     (typical of expressive formations);
frequent presence of the syllable <txi> and its reduplicated form
     <txitxi> (typical of nursery formations);
presence of clusters absent from ordinary lexical items (notably <np>);
very considerable and highly irregular variation in form;
severe localization of each word;
general lack of early attestations."

************************************************************

And here is Trask's challenge:

"Now: does anybody want to make a case that *any* of these words is a good
candidate for native, ancient and monomorphemic status in Basque?"

My answer would be no, not as that question is put.
But they are good candidates for *descendants* from words
which may have had monomorphemic status in ancient Basque.

There is a similar question to which my answer would be yes.

This other question is perhaps not of direct relevance to Trask's
special goals, which he has specified elsewhere,
but I think it is of great relevance to those of us interested in
pushing back towards an understanding of the full vocabulary
of early Basque or pre-Basque.

I would indeed want to claim that the collection of these forms,
taken as a whole, does probably point to the existence of
a small number of ancestral forms in early Basque,
ones which also share some of the properties of words for
'butterfly' elsewhere in Europe.

There are difficulties in reconstructing earlier forms
of many expressives, of two primary kinds.
(1) Parts of them tend to be replaced with sound sequences
which resemble elements of the general vocabulary
or other expressives, thus losing information
about the ancestral forms themselves.
(2) Certain sound patterns may be favored in expressives
of particular meanings, which also can override
ancestral forms.
Both of these can lead to "violations" of sound laws
valid for most of the vocabulary.
But it does not make reconstruction impossible,
it just makes it harder.  Here is why...

An archetypal example of (1) above is the existing (!) English word
<sparrowgrass>, which derives by folk etymology from
<asparagus>.
The analog to the Basque situation would be close if we had
in English dialects also a number of other reformations from
original <asparagus> such as
<parapas>,
<skarakras>,
<askarkras>,
<spargoose>,
<speargoose>,

and so on, all still having the same meaning, the plant 'asparagus'.
How would we reconstruct back from such a set?
With difficulty, obviously, but it is not entirely hopeless.
We might reach a hypothetical
*(a)spara{k/g}(r)as.

Why is this, rather than something else,
a plausible reconstruction?
Forms which contain transparently real word parts
may have undergone greater alterations
as compared with the original,
and forms which contain reduplications or near-reduplications
may have leveled out distinctions existing in the original.
Given these tools of analysis,
not guaranteed valid by any means, but heuristics,
the following are of lesser value for reconstruction,

"sparrow", "spear", "grass", and "goose".

though the words containing them still have some value
since they suggest a
<par> rather than <kar> as more original in the first half,
and  <{k/g}(r)as> rather than <pas> in the second half.
They thus help us to overcome a loss of information
in reduplicated forms.

The first half would then most likely derive from
an earlier <(a)spara...>
and the second half would be something in the range
<pas>
<goose>
<grass>
<kras>
Given the <r> in the first half, the <r> in the second half
might be part of a near-reduplicative effect,
so we might even consider <gVs> most likely.
(Since I derived each of these forms by faily minimal
alterations of known kinds from *asparagus,
and by a number of *different* such alternations,
it is not surprising that reconstructing back from them
we can get something like <asparagus>.)

*************************************************

Consider now what we may be able to do with the Basque data
provided by Larry Trask.
The following techniques and tools are merely heuristics,
they cannot pretend to reach definitive proof,
but I would contend that they are moderately powerful tools,
and they yield a highly plausible hypothesis.
I claim nothing stronger than that,
but that is an achievement not to be merely sneezed at.

I do indeed want to make a case that the majority of
the Basque words cited below are reformations
from an earlier Basque form or forms which
fit the pattern for numerous words for 'butterfly'
at least in parts of Europe (so perhaps with an etymology),
but also elsewhere in the world (so perhaps reflecting
in part some universals of sound-symbolism),
one involving labials, /l/, sometimes /t/, and often reduplication.

<papilon> in Latin would be one of the relatives, and
<butterfly>
<Schmetterling>
<mariposa>
and so on, would also be related.
The idea that these are all ultimately related is not new,
but the difficulties of tracing the details are considerable.

First let us take three of those above, and average their sounds,
removing the <sch-> from the German form:

papi...lon   Latin, cf. French <papillon>
                      violating normal sound changes from Latin to French
                      by preservin the second <p>, regular is <pavillon>.
metter-ling   German, removing initial <sch->, common in
                       other expressive vocabulary
butter-fly    English
                      (I consider English <flutterby> to be a more recent folk
                      etymology, with little value for historical
		      recontruction,
                      just as <sparrowgrass> is recent;
                      probably <butterfly> and certainly <asparagus> are
		      older.)

So we have something like the following,

labial   -   flap-t   -   lateral   -   ?nasal

not as a proven proto-form,
but as an approximate averaging of attested forms.
Since the labial alternative for the second syllable onset
may be a partial reduplication,
it has less value than the other two forms.
Since <butter> is an English word, it has lesser evidential value
than does <metter> in the German,
though there is not much difference between these two in sound.
************************************************************

Now consider Larry Trask's list of Basque dialect words for 'butterfly'.
While I will make comparisons with other European forms,
this implies neither borrowing nor genetic inheritance nor any other
very specific explanation -- it is merely a heuristic to seek
possible explanations.  I will come back later to what seems
to be inferrable from the Basque data alone.

Group 1a.  bitxilote   etc.
Group 1b.  mitxilote   etc.
Group 1c.  txipilota   etc.,  txipilipeta, tximeleta

1a-1b. are very similar to the German and English forms above,

{b/m}itxi...lote

The second stop is Basque <tx> before /i/,
compared with German and English <tt> before a weak vowel.
The third syllable begins with <l>,
and is <lo> as in the Latin and French forms.

The final <...te> has no match in the English, German, Latin-French.
The basque vowels are /i/ instead of the epsilon and caret
vowels of the German and English, or the /a/ of Latin-French.

For 1a, Trask comments:
<bitxi>, western variant <pitxi>, 'pretty little
thing', 'ornament', 'jewel', an item well attested everywhere as an
independent
word (though in varying senses), and also very frequent as a first element in
expressive and nursery formations.

My analysis would be as with <sparrow> and <grass>,
that influence from the word <bitxi> would be late,
and might have pushed the Basque forms
to have medial <tx> instead of   flap-t   or similar,
if that is not ancient,
and to have initial <b...> if that was not ancient but <m...> was.

For 1b, Trask comments:
"These variants show an unexpected initial /m/.  This might result either
from a
perception that they are expressive formations (/m/ is much favored in
expressive formations in Basque), or from contamination by <Mitxel> 'Mike',
the
regular diminutive of <Mikel> 'Michael'.  (Personal names are frequent as
first
elements in expressive names for small creatures: note, for example,
<matxinsalto> 'grasshopper', literally 'Marty-jump'.)"

I would accept the expressive /m/, but consider the
influence of personal names as perhaps quite late.

For 1c.

Following is a quote from Trask:
"These appear to represent metathesized forms of the preceding.  Curiously,
these western forms are entirely absent from Azkue's 1905 dictionary, even
though Azkue was a native speaker of the western dialect Bizkaian, for which
he
provided exceptionally detailed coverage in his dictionary.  Today, the form
<tximileta>, not recorded before 1912, is nearly universal in the western
dialects, and has been accepted as the standard Basque word for 'butterfly'."

"It is especially striking that a form which apparently didn't even exist in
1905 is now the most widespread word in the language."

But there is more to say about  <txipilipeta>.
This has more syllables than any of the others,
and might be a mixing of

<   txipilota>  (attested)
<...pilipeta>  (attested part of the abnormally long form)

If we combine Latin <papilon> with Basque <bitxilote, mitxilote>
we might add to the equation a hypothetical earlier Basque form
with <...ota> instead of the Latin-French <...on>.
Is that at all plausible in terms of Basque or Romance word endings?

**<pipilota>

***

Group 3.

pinpirin (L) (17th c)
pinpirina (L) (17th c)
pinpirineta (Z)
pinpilinpauxa (L) (1905)

These forms have a nasal /n/ in each of the parts of the partial
reduplication, an /n/ which is lacking in the preceding groups.
If we remove it we would have
**pipireta

Thus compare

  <pinpirineta>
**<pi pir  eta> (removing the nasals /n/ and /in/)
**<pi pil  ota> (from the end of discussion of Group 2)
  <pa pil  on>   Latin --> French

Trask comments:
"The Lapurdian dialect is exceptionally fond of expressive formations in
initial
<pin-> and <pan->, a pattern sparsely attested in other dialects; see Lhande's
dictionary of French Basque for more examples.  The last and longest form
appears to contain a palatalized form of eastern <pausa> (n.) 'pause, stop,
hesitation, rest, repose' or its verbal derivative <pausatu> 'pause, stop';
these derive from Latin.  Compare standard Castilian <mariposa> 'butterfly',
literally 'Mary-perch', from <posar> 'perch, alight', itself descended from
the
Latin <pausare>."

I very much suspect that the etymologies using general vocabulary
reflect quite late influences, distorting the earlier expressive word.
Consider the last form, a very long one, which might again represent
a contamination of two forms, and then add in the Spanish <mariposa>

pinpirina
<..pilinpauxa>  (with one nasal /...in/)

  <mari po sa>  (with no extra nasal)

The Spanish "mariposa" in particular, as "mary-perch",
may be a back-formation like "sparrowgrass".
It is too close to <...pilinpausa>, where /m~p/ and /r~l/.
A late contamination with "Maria" is quite plausible.

If the ending <...osa> of <mariposa> is in any way
related to the <...ota> of words discussed above,
we would have

   <papilon>   Latin
   <pipilota>** hypothetical
   <mitxilote>  Group 1
   <mariposa>  Spanish
<...pilinpauxa>  attested in longer form <pinpilinpauxa>
<...pilipeta>  attested in longer form <txipilipeta>

***********************************************

Group 7.  maripanpalona

If we segment this as mari-panpalona, assuming <mari...> as Trask notes,
then we have

<...panpalona
   <pa pilon>    Latin
   <pinpiline>   Group 3

This looks like Group 3,
but with expressive heavy vowels open /a,o/
rather than the light vowel close /i/.

Trask says:
This shows another pattern typical of expressive names for small creatures:
the
use of <Mari> 'Mary' as a first element.  The rest is opaque.  Corominas
suggests a link with Latin <papilio> 'butterfly', but I doubt it.  While I
have
no regional provenance for this form, I suspect that it is eastern, and
eastern
dialects, especially Lapurdian, just love expressive formations in <pin-> and
<pan->, recall.

***********************************************

So far, all of the forms considered may plausibly derive from
a single common proto-form of ancient Basque,
with modifications as is normal for expressive vocabulary.

I make the statement above without attempting to specify
exactly what the ancient Basque form might have been.

***********************************************

The forms of Groups 2 and 6 following may plausibly
derive from another common form of ancient Basque,
one distinct from the one looking like Latin *papilon.
But they may also go back to the same ancient form,
simply with different kinds of changes since then.

Group 2.

txitxidola (LN)
txitxipapa (HN)
txitxitera (Z)

Trask comments:
"These eastern forms exhibit the reduplicated sequence <txitxi>, very common
in
nursery formations, with what appear to be arbitrary final elements: these
final elements have no other existence."

I would agree on reduplication,
but the /...dola, ...tera/  seem metatheses from the /...lota/ of Group 1,
and the /...papa/ seems to reflect the labials so common in words for
butterfly.

***

Group 6.  xintxitoila  etc.

These seem highly similar to the first one in Group 2. --
<xintxi...> is very similar to <txitxi...> ,
perhaps in the way <pipil...> is similar to <pinpilin...>,
so I would assume the same origin for <xintxi...> and <txitxi...>,
whatever that origin was.

Trask comments:
"The first variant [<z...>] is unpalatalized,
while the others show the palatalization typical of expressive formations.
We cannot tell if the first form is conservative or merely a back-formation.
In all its variants, this form is utterly opaque in formation.
The form strongly suggests an expressive
formation particularly typical of the eastern varieties.  See Lhande's
dictionary of French Basque for dozens of examples of this type."

*************************************************

Group 5.

altxatulili
altxa-lili (LN)
altxabili (HN)

The third of these resembles Group 1c.:

al txabili
  <txipilota>  Group 1c.

So I would consider either that this word
may originally not have contained any of the words
/altxa-/, /lili/, or /ibili/,
all words of the normal vocabulary which may have affected
the form of the expressive for 'butterfly' at some point.

Trask comments:
"The first two are transparently compounded from the Romance loan <altxatu>
'raise', stem <altxa->, plus another Romance loan, <lili> 'lily, flower'.  Or
so it would seem, even though the semantic motivation eludes me."

[LA interpolation:  Just as "sparrowgrass" makes no semantic sense.
A good indicator of late reformation.]

Trask continues:
"But the third variant rather muddies the waters.
It may be a somewhat unusual dissimilation of the preceding.
But Agud and Tovar suggest a different formation whose
second element is the common verb <ibili> 'be in motion'.
Maybe, but V-V --> N is a decidedly
unusual type of word-formation in Basque."

*************************************************

So, in sum, the various forms which Trask displays from
Basque dialects do to me suggest at least one reconstructible
ancient Basque word for 'butterfly' (possibly two).

I believe those forms will be something like:

*patilon > *pitxilon / *pitxilota > various
and
*papilon > pinpilin > pinpilin-Vta etc.

Of these two (probably themselves ultimately related),
the first seems more European (see English, German words for
'butterfly'),
while the second seems more specifically like Romance,
and (if we believed that there was some very deep relationship),
I would guess that it is a later borrowing from a relative
of Latin into Basque.  Or could both have been ancient Basque?
The second could derive from the first, by making it
partly reduplicative, more easily than the first could
derive from the second, whether within Basque or
elsewhere or earlier.

The various attested forms result from reduplications,
metatheses, substitution of words from the general vocabulary
which happened to sound much like the earlier form of 'butterfly',
influence from sound patterns common in other expressives, etc.

Despite being normally four syllables,
often with partial reduplication,
it may not be analyzable into morpheme parts.
Does that mean it qualifies as monomorphemic?

***

AFTER writing the above message, I received from Steve Long
a message which he had composed without yet seeing my draft.
He pointed to a wide range of Greek words, among others these:

>ptiloo^
>ptilon (Dor psilon) - soft feathers or down, anything like a wing or feather

and

>"ptil-ôtos, ê, on, winged, esp[ecially] ptilôta, opp[osite of]
>pterôta, dermoptera, membrane-winged creatures, Arist.HA490a6."

[Steve commented:]
>Note the last one - <ptilo^ta>.  Compare Trask's <pitxilota (B)>

to which we can of course add 'wing' *ptero-,

Notice how close this is to what I had reconstructed
from the Basque data,
completely unaware of the Greek
(not even consciously aware of <pteron>,
though I do know the etmology of
<lepido-ptera> and
<ptero-dactyl> when those words are in my consciousness) --
I actually feel a bit silly for not having thought of *ptero-!

Lest someone think this is circular,
the only external hints I used while reconstructing
strictly within Basque were Latin <papilon>,
German <schmetterling>, and English <butterfly>,
oops, also Spanish <mariposa>.

Yet the process led to two hypothetical ancient Basque
proto-forms:

*patilon  (very close to Greek ptilo-)
*papilon

Draw what conclusions you will.

I do not simply conclude that Basque borrowed from Greek,
though that may be most reasonable, since as Roz Frank
points out, there may be no Basque reflex of the
root for 'wing', *ptero- or of *ptilo- other than in
these words for 'butterfly'.

(That very evidence
suggests that it may have been monomorphemic in
early Basque, not transparently divisible into morphological
parts, and thus perhaps relevant to Larry Trask's
search for "monomorphemic" such older items,
despite its multisyllabic character, and even if it was
borrowed at some point in pre-Basque.)

I suppose this etymon must be older than that?
Buck's "Dictionary of Selected Synonyms in
the Principal Indo-European Languages" gives words related to
this under 4.392 'wing' (Greek and OHG, OE) and 4.393 'feather'
(Greek and Germanic and ? Slavic /pero/, assuming that is related.

Nor do I conclude that Basque and PIE are related,
neither from this single piece of evidence,
nor from anything else I have read.
Nor do I conclude that Basque and PIE are unrelated.

***

I look forward to further ideas from either Basque
specialists or specialists in sound symbolism,
or from any other perspectives.
[or, I guess I need to add as an afterthought,
after the Greek evidence has been brought to bear,
from IndoEuropeanists!]

Sincerely,
Lloyd Anderson



More information about the Indo-european mailing list