Pre-Basque lexical items

[Larry Trask]

On Tue, 14 Sep 1999, Jon Patrick wrote:

[LT]

>     Now, my original stricture was against the hundreds of ancient verbs
>     like <etorri> and <ikusi>, whose roots are never free forms.  For the
>     eight or so anomalous verbs like <sartu>, I have no objection if you
>     want to include their stems in your list, since these stems will meet
>     all of my criteria.  Anyway, save only for the anomalous <kendu> (which
>     has other and more regular variants), these stems will in no way be out
>     of line with the forms of non-verbal lexical items generally.

> I don't think I understand this part of your the message (the
> remainder was very informative, thank you). I know I have seen
> <etor> and <ibil> used in the imperative. Are you saying that they
> (and the other <-i> verbs should not be used in their stem form in
> analysis of early basque? It is certainly my intention to use the
> tu/du class of verbs using only their stems.

The ancient verbs, like <etorri> and <ikusi>, are a problem for me,
because my interest is in the morpheme-structure rules of Pre-Basque
lexical items, and the roots of such verbs never occur as free forms.
The morphologically simplest forms, such as <etor> and <ikus>, the
so-called `radical' form of a verb, consist of the stems of the verbs
functioning as free forms.  But these stems invariably contain the
prefix *<e-> preceding the verbal root -- the roots being <-tor(r)-> and
<-kus-> in my examples.

So, I can't use the stems, because these are not monomorphemic: they
always contain that prefix.  But I can't use the roots, either, because
these are invariably word-internal, which produces several
complications.  My solution is to omit these verbs entirely from my
initial list, and to leave them for later consideration.  In any case,
it is already clear that ancient verbal roots are constructed according
to very different rules from those applying to other lexical items, and
I will be interested in characterizing the differences explicitly, so a
separate list for verbal roots is desirable from my point of view.

What you should do about them depends on your purposes.  But bear in
mind that, if you take the stems as data, these stems are bimorphemic
and always contain that prefix.  This is something that will bugger up
the data for some purposes (including mine), though perhaps not for all
purposes.

For the <-tu> class of verbs, the material preceding the suffix <-tu>
does not contain any prefixes and is normally monomorphemic (though not
in all cases).  But, of course, apart from borrowed verbs, which are
numerous and always put into the <-tu> class, almost all <-tu> verbs are
derived from other (non-verbal) lexical items.  Since the majority of
these other lexical items still exist in the language, I presume you
would probably not want to include both the derived verb and its source
word.  For example, you probably wouldn't want to list both <beltz>
`black' and <belztu> `blacken'.  But, for the <-tu> verbs which have no
identifiable source word, such as <saldu> `sell', I don't see any great
problem with listing the stems (<sal-> in this case), though you might
bear in mind that such a stem will often not be phonologically identical
to its source word (witness the possible *<sali> as the source word for
<saldu>).  That is, the stems of such verbs will have undergone the
phonological developments which are regular in Basque word-formation,
such as the loss of a final /i/ in a first element.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk