Lemmatization

Semantic Range, Lexicography, and other approaches to word meaning - in general, or for particular words.
cwconrad
Posts: 2112
Joined: May 5th, 2011, 5:52 pm
Location: Burnsville, NC 28714
Contact:

Lemmatization

Post by cwconrad »

Elsewhere in this forum
RandallButh wrote:... the risk of having people trying to think through rules rather than with words if you give them all of those artificial forms in -αω -οω instead of -ᾱν -οῦν, not to mention needing to signal which voice is preferred?
This raises a serious and very significant question: How should verbs properly be lemmatized?
https://en.wikipedia.org/wiki/Lemma_%28morphology%29 wrote:In many languages, the citation form of a verb is the infinitive: French aller, German gehen, Spanish ir. In English it usually is the full infinitive (to go) although alphabetized without 'to' (go); the present tense is used for some defective verbs (shall, can, and must have only the one form). In Latin, Ancient Greek, and Modern Greek (which has no infinitive), however, the first person singular present tense is normally used, though occasionally the infinitive may also be seen. (For contracted verbs in Greek, an uncontracted first person singular present tense is used to reveal the contract vowel, e.g. φιλέω philéō for φιλῶ philō "I love" [implying affection]; ἀγαπάω agapáō for ἀγαπῶ agapō "I love" [implying regard]). In Japanese, the non-past (present and future) tense is used.
That seems a very accurate account of the convention. Randall several years ago published an article in a Gedenkschrift for Fred Danker, suggesting: (1) using the infinitive as the lemma rather than the 1 sg. present indicative active or middle-passive, (2) preferably using either the present or aorist infinitive depending on which tense is most commonly used, and (3) using either the active or the middle-passive infinitive depending on which voice is more commonly used. That’s consistent, I think, with the notion of using the unmarked rather than the marked form for the verb in question. For my part, I would fully support changing the convention thus. It’s absurd to be using as a dictionary form an uncontracted verb that was never used in ancient Greek, and I think that the clear indication of the more common tense and voice of a verb would be a far better identifier of a verb than the conventional standard.

On the other hand, instituting significant changes in any fixed standard practice is a hard sell -- nigh unto impossible even; my stomach turns when I think of how close the U.S. was to adopting the metric system -- but Congress would not have it (would the Supreme Court have upheld it?).

Do we have any emerging consensus on this issue here? Probably it's easiest to make changes in our own pedagogy, however much we might like to see changes of this sort in major printed reference works. It would surely be much easier to have changes made in electronic lexicographical databases.

What do others think?
οὔτοι ἀπ’ ἀρχῆς πάντα θεοὶ θνητοῖς ὑπέδειξαν,
ἀλλὰ χρόνῳ ζητέοντες ἐφευρίσκουσιν ἄμεινον. (Xenophanes, Fragment 16)

Carl W. Conrad
Department of Classics, Washington University (Retired)
Stephen Hughes
Posts: 3323
Joined: February 26th, 2013, 7:12 am

Efficiency of a system

Post by Stephen Hughes »

cwconrad wrote:Randall several years ago published an article in a Gedenkschrift for Fred Danker, suggesting: (1) using the infinitive as the lemma rather than the 1 sg. present indicative active or middle-passive, (2) preferably using either the present or aorist infinitive depending on which tense is most commonly used, and (3) using either the active or the middle-passive infinitive depending on which voice is more commonly used. That’s consistent, I think, with the notion of using the unmarked rather than the marked form for the verb in question. For my part, I would fully support changing the convention thus. It’s absurd to be using as a dictionary form an uncontracted verb that was never used in ancient Greek, and I think that the clear indication of the more common tense and voice of a verb would be a far better identifier of a verb than the conventional standard.
...
What do others think?
Using such an "umarked" form for learning / teaching, as Randall described, is theoretically efficient. In measuring the efficiency of a starting point in any given system, it is the one with the least distance to all other points in the system, taking into account the frequency of trips to those points is efficient. The most common "trip" in Greek as we read, or tell it to the 3rd person singular aorist indicative active. Choosing the infinitive closest to that makes sense. Having regional depots to suit local conditions makes sense too, when a verb is usually used in the present.

In studying both Classical and New Testament Greek, I have noticed that New Testament dictionaries treat the change in voice as more of a grammatical issue than Classical ones do. I like to list things separately for my own learning and teaching.
Γελᾷ δ' ὁ μωρός, κἄν τι μὴ γέλοιον ᾖ
(Menander, Γνῶμαι μονόστιχοι 108)
Jonathan Robie
Posts: 4237
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: Lemmatization

Post by Jonathan Robie »

cwconrad wrote:On the other hand, instituting significant changes in any fixed standard practice is a hard sell ...
cwconrad wrote:It would surely be much easier to have changes made in electronic lexicographical databases.
Making the data freely available is an important step if we want to make this happen. I would be happy to publish a list of these forms in the biblicalhumanities.org github, if someone has the data. Many of these infinitive forms don't occur in the GNT, which makes it harder to generate such a list automagically.

Does anyone know of a freely licensed list of these forms? Or a list that could be licensed freely?
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/
Stephen Carlson
Posts: 3355
Joined: May 11th, 2011, 10:51 am
Location: Melbourne
Contact:

Re: Lemmatization

Post by Stephen Carlson »

I'm all in favor of lemmatizing based on the infinitive. Research by Shane Kraeger indicates that there was a practice of doing so before the age of printing but at the beginning of which the practice coalesced around the uncontracted first person singular. For a taste of his research, see: https://www.academia.edu/4667773/Histor ... _Testament

As for lemmatizing based on the more common stem/aspect and voice, I'm not so convinced. I feel that the choice of form to lemmatize should be the same, if possible, and that it may not be clear for some verb which is the more common stem. I have played around with the idea that both presents and middles get separate (and with infinitives, usually adjacent lemmas). I suppose that can be done for stem/aspect as well, but I'd rather just put the aorist right after the head word, as if:
λύειν, λύεσθαι
The entry itself can be arranged in order from present to aorist or vice-versa, to reflect which aspect is more dominant, but the choice of lemma itself ought to be fairly predicable to assist word look up. In fact, that's the point of the dictionary, to look up words within it.
Stephen C. Carlson, Ph.D.
Melbourne, Australia
Stephen Hughes
Posts: 3323
Joined: February 26th, 2013, 7:12 am

"Common" is by chance to some degree.

Post by Stephen Hughes »

Stephen Carlson wrote:As for lemmatizing based on the more common stem/aspect and voice, I'm not so convinced.
I also have a concern about frequency. There is an element of historical chance in considering what is common, that is not necessarily a reflection of the whole language. There is of course a logical recessive loop inherent in considering this issue, so not making a choice is a good idea when both forms do occur (at all) in the literature that the dictionary is being made for. When only the aorist, or only the present occur, then the lemma could reflect that.
Jonathan Robie wrote:Many of these infinitive forms don't occur in the GNT ...
As I've inferred regularly, I would prefer to work from a New Testament Greek Dictionary (= A Dictionary of the Greek Language in which the New Testament was written) rather than a New Testament Greek Dictionary (=A Dictionary Containing the words which are in the New Testament). Digitally that could be possible to publish, but to sell it in print with a pitch on the dust-cover something like, "This dictionary contains three times as many words as you are ever going to need to use or learn for your study of the New Testament, weighs an extra 2 kilogrammes and costs an extra $300."

Adequate for the text is easier than "true" to the language, which is something that needs to go beyond the scope of what is currently covered by the best standard works. Moulton and Milligan was at the same time both an God-send (for the understanding of the NT) and an abberation from a dictionary based on a corpus (the papyri) by only choosing the words from the papyri that are used in the New Testament :o .
Jonathan Robie wrote: ... , which makes it harder to generate such a list automagically (sic. :lol: ).
Tom Moore's "Kata Biblon" generates them on demand (unaccented).
Stephen Carlson wrote:The entry itself can be arranged in order from present to aorist or vice-versa, to reflect which aspect is more dominant, but the choice of lemma itself ought to be fairly predicable to assist word look up. In fact, that's the point of the dictionary, to look up words within it.
Do any words occur in the perfect only?
Γελᾷ δ' ὁ μωρός, κἄν τι μὴ γέλοιον ᾖ
(Menander, Γνῶμαι μονόστιχοι 108)
Stephen Carlson
Posts: 3355
Joined: May 11th, 2011, 10:51 am
Location: Melbourne
Contact:

Re: "Common" is by chance to some degree.

Post by Stephen Carlson »

Stephen Hughes wrote:Do any words occur in the perfect only?
ἔοικα is one, I believe. See Jas 1:6.
Stephen C. Carlson, Ph.D.
Melbourne, Australia
Stephen Carlson
Posts: 3355
Joined: May 11th, 2011, 10:51 am
Location: Melbourne
Contact:

Re: Lemmatization

Post by Stephen Carlson »

So in other words, I would:

1. Lemmatize actives and middles separately. So, e.g., the active and middle of λύω get different entries, which helps to reinforce that the middle may have a specialized meaning at times, as with this verb. (This could be a bit too much for middles that are essentially passives of corresponding active intransitives, however.)

2. Lemmatize verbs with a binomial lemma consisting of the present infinitive and aorist infinitive, e.g. λύειν, λῦσαι at the head of one entry and λύεσθαι, λύσασθαι at the head of another.

3. Cross-reference (non-contiguous) aorist infinitives to the present, e.g. λαβεῖν -> λαμβάνειν

4. Within an entry, distinguish and separate the aspect/stems and prioritize based on usage.

As for changes, that's not going to happen unless one writes the reference works oneself. The minimum would be to adopt a practice of referring to the (present) infinitive, so instead of λύω, write λύειν, which is actually a (minority) practice already in use.
Stephen C. Carlson, Ph.D.
Melbourne, Australia
Paul-Nitz
Posts: 497
Joined: June 1st, 2011, 4:19 am
Location: Sussex, Wisconsin

Re: Lemmatization

Post by Paul-Nitz »

Stephen Carlson wrote:So in other words, I would:
I agree with all you have written there.

If it's digital and space is not an issue, I would like to see principal parts listed in each verb entry. If ἑαυτική mid/pass, then a new set of principal parts would be listed. Or, better, I'd like to see what the Lexham Analytical lexicon does, listing every form used in the NT.

Excerpt:
  • Forms of μαθητεύω

    ἐμαθητεύθη VAPI3S(1) μαθητεύω
    μαθητεύεσθαι VPPN(1) μαθητεύω
    μαθητευθεὶς VAPP-SNM(1) μαθητεύω
    μαθητευθῆναι VAPN(1) μαθητεύω
    μαθητεύομαι VPPI1S(1) μαθητεύω
    μαθητεύοντες VPAP-PNM(1) μαθητεύω
    μαθητεύσαντες VAAP-PNM(1) μαθητεύω
    μαθητεύσατε VAAM2P(1) μαθητεύω

    (2011). The Lexham Analytical Lexicon to the Greek New Testament. Logos Bible Software.
Stephen Hughes wrote:Tom Moore's "Kata Biblon" generates them on demand (unaccented).

I couldn't find it on Kata Biblion. Post a link. Is it one-by-one?

I had once, and have again, asked Kalos Beautiful Greek Software designer, G. Diaz, if it is possible to generate a list.
If we started with a dictionary like Abbot, we'd want to do a search and replace, right Jonathon?
Therefore, I told him a CSV list such as the following would be useful. Correct me if I'm wrong.

ποιέω, ποιεῖν, ποιήσαι
ποιέω, ποιεῖσθαι, ποιήσάσθαι

λύω, λύειν, λῦσαι
λύω, λύεσθαι, λύσασθαι
Paul D. Nitz - Lilongwe Malawi
Stephen Hughes
Posts: 3323
Joined: February 26th, 2013, 7:12 am

Re: Lemmatization

Post by Stephen Hughes »

Paul-Nitz wrote:Stephen Hughes wrote:
Tom Moore's "Kata Biblon" generates them on demand (unaccented).


I couldn't find it on Kata Biblion. Post a link. Is it one-by-one?
Yes, it is one-by-one. Of course for most verbs the infinitive is obvious -ein / -sai.
Γελᾷ δ' ὁ μωρός, κἄν τι μὴ γέλοιον ᾖ
(Menander, Γνῶμαι μονόστιχοι 108)
Stephen Hughes
Posts: 3323
Joined: February 26th, 2013, 7:12 am

Sounds like a good plan

Post by Stephen Hughes »

Paul-Nitz wrote:Stephen Carlson wrote:
So in other words, I would:

I agree with all you have written there.
Sounds like a good plan.
Γελᾷ δ' ὁ μωρός, κἄν τι μὴ γέλοιον ᾖ
(Menander, Γνῶμαι μονόστιχοι 108)
Post Reply

Return to “Word Meanings”