Q: `lemma'

[Larry Trask]

----------------------------Original message----------------------------
 
Me again.  This time I'm asking about the use of `lemma'.  The
general sense of this seems to be something like `a linguistic form
heading some discussion or accompanied by an explanation', but in
practice usage varies somewhat.  At the moment, I have four distinct
definitions, as follows, all of them attested somewhere:
 
1. A headword in a dictionary.
 
2. A reconstructed or attested form which serves to head a list of
linguistic forms which are descended from it.
 
3. A historically attested form representing a reconstructed form.
 
4. A cited linguistic form accompanied by a gloss.
 
Senses 2 and 3 are perhaps slightly conflicting.  For example, Calvert
Watkins, in his dictionary of PIE roots, quite explicitly defines a
lemma as a historically attested form continuing a reconstructed root,
and cites as an example the entry for PIE *<ekwo-> `horse',
represented by (1) Latin <equus> and (2) Greek <hippos>, in which the
Latin and Greek words are for Watkins the lemmas (lemmata).  Here I
would have said that *<ekwo-> was most obviously the lemma, though I
can see the point of applying the term also to each of the
subheadings, but I'm puzzled about the statement that a lemma is
"historically attested".  And, of course, in many entries some of the
enumerated subheads are themselves reconstructed, not attested, as for
example when PIE *<gwou-> `ox, bull, cow' is followed by (1)
Proto-Germanic *<ko:uz> and (2) Latin <bo:s>, among others.  By
Watkins's definition, <bo:s> is a lemma here but *<ko:uz> is not,
which appears to make very little sense.
 
On definition 4, it is not clear to me whether the intention is to
include only roots, stems and lexical items, or whether inflected
forms should also be included.  So, for example, if I cite Basque
<ikusi> `see', then <ikusi>, a lexical item in its citation form,
would be a lemma (at least for some people), but what about its
inflected form <dakusat> `I see it'?  Is <dakusat> here also a lemma?
 
All comments gratefully received.
 
Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK
 
larryt at cogs.susx.ac.uk