Computer Aided Language Learning - Principal Parts

[Jonathan Robie]

Suppose you wanted to design software to teach verb morphology, including the principal parts. What would this software look like?

Or would you use the software to generate worksheets that people could use?

So far, I'm convinced that:

1. Flashcards based on tables aren't that helpful, at least for me. Playing with the Moore College Anki deck, which is very well defined, I find that I may not be able to recall a form in the principal parts table unless I think of the word in a sentence.

2. Reading sentences in context is probably more helpful, and I could definitely create a query that would generate a list of examples of verbs, grouped by verb, then grouped by each principal part of the verb. One relatively easy solution would be to generate a catalog of examples people could print out and play with.

Any thoughts on the best way to leverage the corpus to help people learn verb morphology?

[cwconrad]

Suppose you wanted to design software to teach verb morphology, including the principal parts. What would this software look like?

Or would you use the software to generate worksheets that people could use?

So far, I'm convinced that:

1. Flashcards based on tables aren't that helpful, at least for me. Playing with the Moore College Anki deck, which is very well defined, I find that I may not be able to recall a form in the principal parts table unless I think of the word in a sentence.

2. Reading sentences in context is probably more helpful, and I could definitely create a query that would generate a list of examples of verbs, grouped by verb, then grouped by each principal part of the verb. One relatively easy solution would be to generate a catalog of examples people could print out and play with.

Any thoughts on the best way to leverage the corpus to help people learn verb morphology?

I spoke briefly with Jonathan yesterday afternoon about this and I've been thinking about it overnight. I need to do some more thinking but a couple thoughts have come to mind.

One thought was triggered by Tim Nelson's comment that it was spending sufficient time grasping principles of Greek word-formation that consolidated his grip on morphology. I am wondering if we (not I, of course) could conjure up a program that would display the formative elements of numerous forms of the same verb in sizable syntactic units within the verse-context of their appearance in the GNT, such that each formative element shows in its own color. For a quick illustration, I have in mind something like this: καταλλαχθησομένων. It's obvious that I don't know how to use the colors to best advantage, but my idea ought to come through: one needs to sense the presence of these formative elements in these complex Greek verb forms while at the same time seeing the verb forms in meaningful contexts within a text. I think that one could work up groups of GNT texts showing all the instances of each of the irregular verbs and tie them together with an explanatory account of how these formative elements function in the morphology of the Greek verb -- including how tense-stems are formed from verb roots and how phonetic principles govern linkage of formative elements (e.g. -θη-σ-ο-μέν-οις but θε-ντ-ι, κατ-αλλαγε-ντ-ος but κατ-αλλαγε-ντy-ης = καταλλαγείσης).

I think that the explanatory account would have to show how the irregular verbs form tense-stems in each of the major tense-systems and how the monosyllabic or disyllabic roots alter in combination with single consonants or consonant clusters. Basic Greek phonology would have to enter into that, but an additional problem with regard to phonology is that the morphological construction of Greek verbs of the NT Koine is based upon much older rather than upon current pronunciation.

There are other questions needing answers before this sort of project could be carried out. It seems to me that this sort of program would never be needed for one who learns Greek through conversational interaction rather than through study that involves primarily reading printed texts. One who learns Greek fundamentally through hearing and speaking doesn't need to deal with grammar at all -- or very little, but of course, reading is largely a different skill. One who learns Greek fundamentally through hearing or speaking doesn't need to learn principal parts (although he or she might get hung up over "hung" and "hanged" or "lay" and "laid").

[jtauber]

I am wondering if we (not I, of course) could conjure up a program that would display the formative elements of numerous forms of the same verb in sizable syntactic units within the verse-context of their appearance in the GNT, such that each formative element shows in its own color.

I have no sense of the best way to present it, but I'm getting close to having the raw data to enable this.

[jtauber]

I am wondering if we (not I, of course) could conjure up a program that would display the formative elements of numerous forms of the same verb in sizable syntactic units within the verse-context of their appearance in the GNT, such that each formative element shows in its own color.

I have no sense of the best way to present it, but I'm getting close to having the raw data to enable this.

Incidentally, one thing I'm trying to make sure my data captures is where the principal parts are predictable from each other as well so just the irregularities can be extracted (or tested differently) if desired.

[Jonathan Robie]

James and I are working on different halves of this problem, and both are important.

I wrote an XQuery that returns examples for the forms of a verb, organized around the principal parts, by number and person within each part. Here's output for the verb λέγω. I can improve this over time, and expand this to create separate output for each verb in a list, but I hope this gives the basic idea.

Now ... imagine this with the morphemes broken out and colored, as Carl and James have been suggesting. And perhaps with some introductory documentation here and there, such as the principal parts for a given verb. Is this a good way to get more experience playing with the morphology of a verb in context?

For instance, if you have a list of examples like this, it's fairly easy to change the examples, e.g. change who is saying what to whom, change the tense, etc., using examples drawn from real text. You can do that as a worksheet exercise, writing Greek, or it could be automated using cloze or whatever ... perhaps that would be more efficient than memorizing principal parts? Or perhaps it could at least work for those who don't do well learning form tables and don't have a live teacher to work with them ...

[Stephen Hughes]

I can't exactly remember the process I went through to learn the principal parts, by I do know that in reading the association between one part or other and the dictionary form is most difficult for suppletives, quite difficult for those where the root is modified and easy for most verbs.

Can your generator generate lists sorted by obviousness?

[Jonathan Robie]

I can't exactly remember the process I went through to learn the principal parts, by I do know that in reading the association between one part or other and the dictionary form is most difficult for suppletives, quite difficult for those where the root is modified and easy for most verbs.

Can your generator generate lists sorted by obviousness?

If we define obviousness in some way, yes. And that can be done individually for each verb.

Here's a directory of some verbs, with links to overviews of each verb, including all the common suppletives, among others. Easy enough to generate with a query ... what's the next step?

Is this starting to look promising, at least for teachers thinking of how to design exercises for students, or very motivated students for self-learning?

[Jonathan Robie]

There are other questions needing answers before this sort of project could be carried out. It seems to me that this sort of program would never be needed for one who learns Greek through conversational interaction rather than through study that involves primarily reading printed texts. One who learns Greek fundamentally through hearing and speaking doesn't need to deal with grammar at all -- or very little, but of course, reading is largely a different skill. One who learns Greek fundamentally through hearing or speaking doesn't need to learn principal parts (although he or she might get hung up over "hung" and "hanged" or "lay" and "laid").

This is certainly not a replacement for a conversational approach - but even someone preparing classes using a conversational approach benefits by easy access to lots of example sentences. And these can serve as a framework for writing exercises, which are also a useful "living language" approach.

For what it's worth, the principal parts aren't explicitly mentioned in the verb explorer, not even in the group titles. But tense, person, and number are mentioned in the group titles.

[cwconrad]

\Here's a directory of some verbs, with links to overviews of each verb, including all the common suppletives, among others. Easy enough to generate with a query ... what's the next step?

Is this starting to look promising, at least for teachers thinking of how to design exercises for students, or very motivated students for self-learning?

Interesting. I may not have explored far enough, but it looks like verbs are classified for voice only as active or passive: ἀποκρίνομαι, for instance, and the participles ἀποκριθείς are both listed as passives. Maybe I'm misreading it, but if that's the case, I like it; it seems to me to match Dionysius' identification of verb forms as ἐνεργητική and παθητική, μεσότης being used for verbs that have forms in both morphological patterns.

[Alan Bunning]

I am wondering if we (not I, of course) could conjure up a program that would display the formative elements of numerous forms of the same verb in sizable syntactic units within the verse-context of their appearance in the GNT, such that each formative element shows in its own color.

I have no sense of the best way to present it, but I'm getting close to having the raw data to enable this.

Incidentally, one thing I'm trying to make sure my data captures is where the principal parts are predictable from each other as well so just the irregularities can be extracted (or tested differently) if desired.

I have data and a program that can generate word forms along the lines that are being discussed. It is fed a list of roots and only the irregular forms that are not predictable and then uses the normal rules for augments, endings, and contractions. Because it does it this way, it could easily colorize the different components. However, my program was designed to be used in reverse – for checking whether parsings are correct. Given a word and its parsing, it generates what it thinks the expected word should be, and then compares that with the word it is given. If the words don’t match, I then go back and check to see if the parsing is correct. Given every form of every word in the New Testament, there are only about 300 word forms that don’t match what my program thinks they should be. And in those cases, they just don’t follow the rules I gave them, so the program would just have to be told. (Either that, or I would have to think up some more rules that might handle some of those exceptions.)

I used this program to generate the aorist infinitive forms of all of the New Testament words for Jonathan. But that raises another issue in that such a program can generate a word form, and it could even be correct, but that does not mean that such a word was ever actually used in the literature. There are other complexities to this process that I won’t go into here, but the whole thing was not as easy as you might think. For example, some words are conjugated normally, and then also have an irregular form for the same parsing. And then when you consider all of the weird spellings across all of the manuscripts I have collected...

Unfortunately, that project was put on hold as I am bogged down right now with finishing up my electronic transcription of Codex Ephraemi Rescriptus and am also trying to put all of my data in a format for exchange to make Jonathan happy. I would be happy to collaborate and share data though, as maybe there would be some synergy in working with Jonathan and James on this issue.