[B-Greek] Lemmatization (was "Inaccurate electronic versions")

Carl W. Conrad cwconrad at artsci.wustl.edu
Thu Jan 25 09:22:44 EST 2007


On Jan 25, 2007, at 6:26 AM, James Tauber wrote:

>
> It's the lemmatization that I've mostly focused on over the last
> twelve years on and off.

Lemmatization and parsing are more problematic than is commonly  
realized, I think, particularly with respect to verb-forms.   
Lemmatization in lexica and parsing tags in parsing guides could be  
altered in ways that would better reflect the real structure of the  
ancient Greek verbal system especially.

One hesitates to depart from the convention of listing verbs  
preferably by the first-person singular present indicative active,  
even if an active-voice form or a present-tense form is either rare  
or non-standard.
It's customary, for instance, to list contract verbs in the  
uncontracted form of the present indicative 1 sg. -- despite the fact  
that one doesn't ever see the uncontracted forms (outside of old  
Ionic). The rationale for this is that knowing the stem vowel is -A-,  
-E-, or -O- assists one consulting the lexicon to recognize  
contracted forms of the verb (provided that one really knows the  
contractions, of course).
Two changes in standard lemmatization of verbs have been suggested:
(1) Infinitives, rather than 1 sg. indicative forms, might better  
characterize the verb, e.g. LEGEIN instead of LEGW, POIEIN instead of  
POIEW. POREUESQAI instead of POREUOMAI.
(2) Randall Buth (and perhaps some others) has suggested that the  
aorist-tense form should be used as a lemma instead of a present- 
tense form, in which case we'd have ELEXA (or better LEXAI,  
especially since the infinitive is unaugmented) for LEGW, EPOIHSA (or  
POIHSAI) for POIEW, EPOREUSAMHN (or better POREUSAI) for POREUOMAI. I  
personally think that using the aorist infinitive instead of the  
present-tense indicative 1st sg. would be a more useful lemmatization.

Much of my own focus of though in recent years has been on voice  
morphology and designations most appropriate for the standard  
categories and of voice-form notation:

(1) Middle-passive and Passive morphoparadigms: If we can rid  
ourselves of the useless and misleading notion of "deponency" and  
recognize that the -QH- and -H- aorists (conventionally termed "first  
passive" and "second passive") are identical in form with non- 
thematic second aorists ("active") such as ESTHN and EFANHN, and  
grasp that, beginning already in Homeric poetry, -QH- aorists  
increasingly over the centuries replaced the older aorist Middles in - 
OMHN, etc. while retaining both middle and passive semantic force,  
then we should distinguish in our parsing (a) the Middle-passive  
forms in -MAI/SAI/TAI, MHN/SO/TO from (b) the Middle-passive forms in  
-QH- (including futures in -QHSOMAI and -HSOMAI). The question is how  
best to do this; I have suggested in the past that we should  
designate the MAI/SAI/TAI, MHN/SO/TO forms as "MP1" and the -QH-  
forms as "MP2" (this would properly indicate that any of these forms  
might potentially carry either "middle" or "passive" semantic force  
-- although whether it is "passive" rather than "middle" really  
depends upon the particular verb in question). If, however, adoption  
of these designations ("MP1" and "MP2") seems too radical, then we  
might retain the more traditional markers "M" for the MAI/SAI/TAI,  
MHN/SO/TO forms and "P" for all the -QH- forms. Of course, changes  
such as this would require a "paradigm shift" in pedagogy with  
respect to the teaching of the voice-system of ancient Greek (it was  
not I but Neva Miller who first spoke of such a pedagogical "paradigm  
shift").

(2) Middle verbs and verbs for which the Middle form is standard  
(e.g., hISTAMAI, EGEIROMAI): Perhaps as 'radical' as using the aorist  
infinitive as the primary lemma for verbs would be a shift to  
indicating as lemma the middle form (present indicative active 1 sg.  
or aorist infinitive) of verbs that are fundamentally middle and  
intransitive  despite the fact that they have active causative forms.  
So-called "deponents" even now are lemmatized with middle-forms  
(POREUOMAI, DUNAMAI, ERCOMAI), but there are several verbs (termed  
"ergative" by some linguists) that are fundamentally intransitive and  
have middle-voice forms although they have transitive active- 
causative forms: hISTAMAI "come to a stand" and EGEIROMAI "rise" are  
examples; their corresponding active forms are hISTHMI "make to  
stand" and EGEIRW "make to rise/raise." There are also some less- 
frequently-used verbs that are pretty clearly middle but are  
occasionally used in active forms: AGALLIAOMAI, for instance, appears  
eleven times in the GNT, 9x in the middle-passive or passive, 2x in  
the active; clearly the middle-passive forms are standard, yet the  
active form AGALLIAW is the lemma in the lexica: it ought not to be.

>
> You are correct that the original CCAT database made available in
> 1993 has a lot of errors -- thousands, if you count instances.
>
> If you get the CCAT file from anywhere other than jtauber.com or
> morphgnt.org, you are probably getting either the original CCAT
> release from 1993 or one from around 1995 with my first round of
> corrections. It's the latter which Broman and others were making
> available (and which was the basis for the late Tony Fisher's site).
>
> However, many significant corrections have been made since then,
> especially in 2002 and since 2004.
>
> Since December 2005, I've been working with Ulrik Petersen (who in
> parallel to my CCAT/UBS work had been working on Tischendorf and
> other texts). We are working on integrating our data, the first
> fruits of which should be available soon.
>
> James
>
>
> On 24/01/2007, at 6:35 PM, Alan Bunning wrote:
>
>> Thanks for the info! I'll take a look and compare it with what I
>> have. I
>> certainly don't want to have to reinvent the wheel. It does look
>> like there
>> is some overlap. Let me know if anything I am doing can help your
>> cause. The
>> next thing I was planning on doing, was correcting the parsings and
>> roots
>> for those same texts. As I said, I barely got started and already
>> found a
>> tremendous number of errors. What sort of work have you done in
>> this regard?
>> I will certainly add the set you have to the mix when I compare them.

Carl W. Conrad
Department of Classics, Washington University (Retired)
1989 Grindstaff Road/Burnsville, NC 28714/(828) 675-4243
cwconrad2 at mac.com
WWW: http://www.ioa.com/~cwconrad/





More information about the B-Greek mailing list