BWS Clause Participant question relating to πλοῖον

Page 1 of 1 (11 items)
This post has 10 Replies | 2 Followers

Posts 22020
Forum MVP
Graham Criddle | Forum Activity | Posted: Tue, Aug 18 2015 9:41 AM

The screenshot below shows:

  • on the right a couple of expanded sections from the BWS for πλοῖον (boat). This is in the Clause Participants -> "In Relation to" section
  • on the left a Passage List of references I would have expected to be included in the BWS report

There is one other "in relation to" section further down that refer to Jesus ("Boat of Jesus' teaching") but the references in that section are already included in the top one shown in the screenshot.

Are there reasons why the missing references aren't included or is this an omission in tagging?

Thanks, Graham

Posts 86
Scott Fleischman | Forum Activity | Replied: Mon, Aug 24 2015 5:21 PM

I think what is happening here is that Clause Participants "In Relation To" section shows cases where both the lemma and the biblical entity occur in the same verb argument. For πλοῖον and Jesus, that will usually happen when there's a subordinate clause "ἐμβὰς εἰς πλοῖον" that is an argument to the main verb "διεπέρασεν".

If you look at the syntax of Mt 14:13, ἐν πλοίῳ is ADV to ἀνεχώρησεν and there is no argument that includes both Jesus and πλοῖον.

Posts 22020
Forum MVP
Graham Criddle | Forum Activity | Replied: Wed, Aug 26 2015 6:43 AM

Hi Scott

Thanks for getting back to me - but I'm still not quite clear about this

Scott Fleischman:
I think what is happening here is that Clause Participants "In Relation To" section shows cases where both the lemma and the biblical entity occur in the same verb argument

Interested to understand why it is limited in this way and doesn't span the entire clause

Scott Fleischman:
If you look at the syntax of Mt 14:13ἐν πλοίῳ is ADV to ἀνεχώρησεν and there is no argument that includes both Jesus and πλοῖον

True - but that seems to be the case for Matthew 8:23 as well which is returned in the results

Posts 86
Scott Fleischman | Forum Activity | Replied: Wed, Aug 26 2015 9:56 AM

Graham Criddle:
True - but that seems to be the case for Matthew 8:23 as well which is returned in the results

Jesus is the αὐτῷ in "ἐμβάντι αὐτῷ εἰς πλοῖον" which is a subclause of the verb ἠκολούθησαν. Since both Jesus and πλοῖον are in the same argument (the subclause argument of ἠκολούθησαν), the verse gets returned in the results. In Mt 14:13 there is no subclause that contains Jesus and πλοῖον.

I double checked the code, and it is doing what I suspected. For non-verb lemmas, It's finding biblical entities that are in the same verbal argument as the lemma. Since subclauses are arguments, it also finds clauses that contain both the lemma and the entity anywhere in that subclause. It does return good results for those subclauses so one might wonder why it doesn't do so for main clauses.

Graham Criddle:
Interested to understand why it is limited in this way and doesn't span the entire clause

That's a good question, and I don't have a good answer. Maybe Eli can chime in on why it is the way it is.

I think, though, that the results would become noisier if the match spans the entire clause because of subclauses. If the biblical entity were in the main clause, and the lemma buried in a subclause somewhere, the results could look incorrect because of the distant "relationship" between the entity and the lemma.

Posts 22020
Forum MVP
Graham Criddle | Forum Activity | Replied: Wed, Aug 26 2015 10:38 AM

Scott Fleischman:

Graham Criddle:
True - but that seems to be the case for Matthew 8:23 as well which is returned in the results

Jesus is the αὐτῷ in "ἐμβάντι αὐτῷ εἰς πλοῖον" which is a subclause of the verb ἠκολούθησαν. Since both Jesus and πλοῖον are in the same argument (the subclause argument of ἠκολούθησαν), the verse gets returned in the results. In Mt 14:13 there is no subclause that contains Jesus and πλοῖον.

That's reasonable - although I admit it wasn't immediately obvious to me that "he got into the boat" in Matt 8:23 was a subclause of "his disciples followed him". 

Scott Fleischman:
I double checked the code, and it is doing what I suspected. For non-verb lemmas, It's finding biblical entities that are in the same verbal argument as the lemma. Since subclauses are arguments, it also finds clauses that contain both the lemma and the entity anywhere in that subclause. It does return good results for those subclauses so one might wonder why it doesn't do so for main clauses

Thanks for checking and the clear explanation - really appreciated.

Scott Fleischman:

Graham Criddle:
Interested to understand why it is limited in this way and doesn't span the entire clause

That's a good question, and I don't have a good answer. Maybe Eli can chime in on why it is the way it is.

That would be great

Scott Fleischman:
I think, though, that the results would become noisier if the match spans the entire clause because of subclauses. If the biblical entity were in the main clause, and the lemma buried in a subclause somewhere, the results could look incorrect because of the distant "relationship" between the entity and the lemma.

That's possibly true. But the problem at the moment is that I am trying to find instances where Jesus appears in relation to a boat (and this is just the example I am using to demonstrate!) and there are some which appear "obvious" in English but are not picked up by this section of the guide. As such I need to use other search methods to check the results. 

And it seems somewhat reasonable to expect a section within "Clause Participants" to return information at the clausal level.

And in the Matt 8:13 case I don't think there are subclauses, just a number of Adverbial Functions within a single clause.

Posts 791
LogosEmployee
Eli Evans (Faithlife) | Forum Activity | Replied: Wed, Aug 26 2015 10:59 AM

Graham Criddle:

Graham Criddle:
Interested to understand why it is limited in this way and doesn't span the entire clause

That's a good question, and I don't have a good answer. Maybe Eli can chime in on why it is the way it is.

When you run a BWS on a verb, it's fairly obvious what you would want the grammatical and semantic roles to list: All the entities that fulfill a given grammatical or semantic role as an argument of the given verb. There have been quirks in the implementation and data, but at least it's pretty clear what the "right" answer is. It's also pretty clear which clause the answers should be drawn from, because the given verb defines the clause.

But we struggled to find anything useful to display for non-verbs. Would you want to see every entity that co-occurs within the same clause? That's a possible right answer, but it generates a lot of noise. The longer the clause, the less useful it was to see that entity X appeared in the same clause as given lemma Y.  

(It sounds like maybe this is what you were expecting.)

It got worse when we listed what constituent the found entity was in: So we found "boat" in the object of a clause, so we listed "Jesus" under "Subject." And then we found "boat" and "Jesus" together in another subject (in an embedded clause), so we listed "Jesus" under "Subject." But the two things aren't very much alike, and worse, "Jesus" is never the subject of "boat" -- that's obvious nonsense.

So we thought to limit it to the same constituent of the clause at least. That seemed much better. Again, we tried to sort the results by the grammatical or semantic role of the containing constituent. And it still looked like "Jesus is the subject of boat" nonsense.

At one point I despaired and thought we just shouldn't show anything for non-verb lemmas.

In the end, we fell back to something that was vague enough to be correct for all the cases we found: x in some unspecified relation to y. The relation is "occurs in the same clause constituent" and the nature of the relationship is unknown.

Clearly this doesn't find every imaginable relationship. If we were to craft queries that would find all the relationships you could imagine, then we'd end up with a lot of noise to boot. It's about striking a balance between signal and noise, and ... for better or worse, this is the balance we struck.

Posts 22020
Forum MVP
Graham Criddle | Forum Activity | Replied: Wed, Aug 26 2015 12:54 PM

Hi Eli

Many thanks for this detailed response - really helpful

Eli Evans:
but at least it's pretty clear what the "right" answer is. It's also pretty clear which clause the answers should be drawn from, because the given verb defines the clause

Agreed

Eli Evans:

But we struggled to find anything useful to display for non-verbs. Would you want to see every entity that co-occurs within the same clause? That's a possible right answer, but it generates a lot of noise. The longer the clause, the less useful it was to see that entity X appeared in the same clause as given lemma Y.  

(It sounds like maybe this is what you were expecting.)

Yes - this is what I was expecting but the insight you give as to the issues that causes helps me understand something of the problem.

Eli Evans:
In the end, we fell back to something that was vague enough to be correct for all the cases we found: x in some unspecified relation to y. The relation is "occurs in the same clause constituent" and the nature of the relationship is unknown.

Eli Evans:
for better or worse, this is the balance we struck.

And understanding this is really helpful.

Appreciated, Graham 

Posts 570
HansK | Forum Activity | Replied: Wed, Aug 26 2015 3:02 PM

Graham Criddle:

Graham's search shows 9 hits in the BWS report (clause recipients). When I do the same search on all passages, I only get 6 results:

What is wrong here (or what I am doing wrong)? I have the latest beta installed and work on a Mac.

Hans

MacOS Sierra / Logos 7 Collector's Edition & All Base Packages / Logos Now

Posts 22020
Forum MVP
Graham Criddle | Forum Activity | Replied: Wed, Aug 26 2015 3:15 PM

You're not doing anything wrong.

There was an update (resource as opposed to code I think) since the time I posted my original screenshot and now

Posts 570
HansK | Forum Activity | Replied: Wed, Aug 26 2015 3:29 PM

Thanks for replying. So you have 6 results now too ?

Big Smile

MacOS Sierra / Logos 7 Collector's Edition & All Base Packages / Logos Now

Posts 22020
Forum MVP
Graham Criddle | Forum Activity | Replied: Wed, Aug 26 2015 11:34 PM

HansK:
Thanks for replying. So you have 6 results now too ?

That is correct

Page 1 of 1 (11 items) | RSS