Excluding data: Azkue's dictionary

[Larry Trask]

Before we abandon this thread, I might make a few comments on Azkue's
1905 dictionary of Basque, which Jon Patrick is using to set up his
database.

This dictionary offers some advantages as the source of a database, but
also a few drawbacks.

On the plus side, it is pretty comprehensive.  Very few Basque words
recorded at all before 1905 are missing from it.  It includes virtually
all words used in writing before Azkue's compilation; it includes a
number of words reported by linguists and lexicographers even though
they are absent from the written texts; and it includes a sizeable
number of words collected by Azkue himself in his investigations.
Until the completion of the Basque Language Academy's new dictionary,
Azkue's will remain the most nearly comprehensive Basque dictionary we
have.

Moreover, Azkue is punctilious about recording provenances.  Not only
for each word, but for each distinct sense of a word, Azkue records the
region in which, as far as he knows, the form and meaning are attested.
This is valuable information.

A further virtue is that Azkue enters bound morphemes, something that
few Basque dictionaries do -- though Azkue's accounts of these are not
always satisfactory.

But there are also major drawbacks.

First, the dictionary contains a number of errors.  Some of these are
ghost words resulting from typos, from Azkue's mishearing of a local
word, or from his misunderstanding of an inflected form in print.  These
errors number at least some dozens, though most of them were later noted
by Michelena in his article on the dictionary and in other writings.
These errors really need to be corrected in any large-scale use of the
book.

Second, Azkue, who rather fancied himself as an etymologist, included
entries for a fair number of non-existent words, of which the most
infamous is his putative *<iz> `water', which has probably been seized
upon by more delighted comparativists than any single genuine Basque
word.  These little fantasies are never overtly marked as such, though
most of them are identifiable by some such annotation as "now reduced to
a radical" or "no longer in use as an independent word".  Any entry with
such an annotation should be discarded.

Third, when -- as so often -- a word exists in several regional
variants, Azkue enters each variant separately in its own alphabetical
place, and he hardly ever provides cross-references.  Consequently, any
attempt at using his dictionary without heavy editing will lead to
multiple entries for single words.  This is potentially a serious
problem: for example, the Basque word for `strawberry' exhibits at least
twenty regional variants.  This is admittedly an exceptional case, but
very many words exhibit two to six regional variant forms, and accepting
Azkue's headwords without suitable editing will produce badly skewed
results.

Fourth, Azkue generally provides only a single headword for homophones
of unrelated meaning.  This too will skew the results, though in a
different direction from the preceding.

Finally, of course, Azkue makes no attempt at reporting dates of first
attestation -- a grievous shortcoming for certain purposes, including
mine.

These shortcomings add up to potentially formidable difficulties in
compiling a useful database from its headwords.

I suggest that a much better dictionary for the purpose is Sarasola's
1996 dictionary.  This is not so comprehensive as Azkue, but the words
excluded are those which are only marginally reported anyway.  In
contrast to Azkue, Sarasola provides dates of first attestation, and he
also gathers regional variants under a single headword, with
cross-references where necessary.  Moreover, he lists the known senses
in the order in which they are attested -- a very valuable feature for
historical work.  I consider it unlikely that many attested Basque words
which are genuinely native and ancient are excluded from this
dictionary.  The sole obvious shortcomings are that Sarasola is much
less explicit than Azkue in recording the regions in which each word is
attested, and that the book does not enter bound morphemes.

I suggest that, for any study of ancient Basque words, Sarasola's
dictionary is a better choice than Azkue's, though it might be usefully
supplemented by the information in Azkue, especially for provenances.
But, of course, Sarasola's dictionary is written entirely in Basque,
which is not very convenient if you can't read Basque.  Azkue's, in
contrast, is Basque to Spanish and French.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk