Help with Syntax Search

Page 2 of 2 (27 items) < Previous 1 2
This post has 26 Replies | 0 Followers

Posts 25223
Forum MVP
Graham Criddle | Forum Activity | Replied: Fri, May 20 2011 7:01 AM

Gregory B. Smith:
 Which is the best?

I see it as less that there is a "best" - it is rather that they do different things.

http://community.logos.com/forums/p/3389/26326.aspx#26326 is a (old) post from Rick which summarises the key differences (the Lexham resource is now complete and available although it wasn't when Rick wrote the post)

Graham

Posts 19
Houston B Smith | Forum Activity | Replied: Fri, May 20 2011 7:51 AM

After reading your earlier post, I think I understand.  You are viewing Δαυιδ as the object of the preposition and both οικος and πατρια as genitives defining Δαυιδ.  Is that correct?

Posts 25223
Forum MVP
Graham Criddle | Forum Activity | Replied: Fri, May 20 2011 8:25 AM

Hi Gregory

Gregory B. Smith:
After reading your earlier post, I think I understand.

I think this refers to the post made by Rich - and I'll leave it to him to respond!

Can I encourage you to look at using the "Quote" feature when replying to a post to make it easier to keep track of conversations?

Details can be found under the section entitled "Reply Post Editor" under http://wiki.logos.com/Using_the_Forum#Adding_a_Reply_to_a_post

Graham

Posts 1419
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Fri, May 20 2011 9:20 AM

Gregory B. Smith:

After reading your earlier post, I think I understand.  You are viewing Δαυιδ as the object of the preposition and both οικος and πατρια as genitives defining Δαυιδ.  Is that correct?

I'd say that's how the Cascadia Analysis presents it. :)

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 1419
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Fri, May 20 2011 11:41 AM

Rick Brannan (Personal):

Hi Gregory.

Gregory B. Smith:
It found a number of results, but I am trying to understand why it missed Luke 2:4

You're looking for prepositional phrases where a single preposition has multiple prepositional objects. In Lu 2.4, the preposition only has one object (Δαυίδ); that noun is modified by a compound phrase (οἴκου καὶ πατριᾶς).

Now, if you wanted to, for Phrase 2 (the containing phrase object of the compound nominal phrase) you could set the "is direct child of" property (don't remember the exact terminology in the interface) to false. That would find the compound structure within the prepositional phrase hierarchy. It might also find a bunch of other stuff you may not want, though.

I'll see about some further queries / etc. when I get into the office. I might have some other approaches that could help or at least stimulate thinking.

This is the follow-up I mentioned earlier. The search as originally constructed would also miss out on valid (according to Cascadia) compound-object prepositional phrases like Mt 3.11:

Here the object has an adjectival phrase further modifying it, but the first query didn't locate it because of the strict np->np->noun structure. You can get around that by using the syntax query editor "agreement' capabilities. First, a digression.

Cascadia is an example of (I think this is the correct terminology) head-driven phrase grammar. So in the Cascadia graphs, when a node has more than one child, an asterisk notes which node is considered the head of the structure. When there is only one child, many times (not all the time) the head is not specifically noted by an asterisk, but there is a 'head' nonetheless. The data for the 'head' is encoded at each node. In this case the 'head' of the CL is the verb (finite verbs are almost always the clause heads by definition) so the data for βαπτισει is encoded at the CL, the V, the vp and the verb. You can see it in the graphs just by hovering over the node (look to the bottom left corner of the window, at least in the Windows version of Logos4). This allows that information to be accessed anywhere along the graph when searching. We also encode 'instance' information at each point, so it can be known which word 'instance' the head actually points to.

Because of all of this, we can match a node (like a noun phrase (np)) to its 'instance' using agreement and 'matching skips levels', and this means we don't have to worry about the levels between (the levels are what causes Mt 3.11 not to match the original query).

Apologies because I'm using 4.3 beta 5, which has a bug that prevents 'detail' view of the query, and only shows 'concise' view. But here's a query with agreement:

The 'terminal node' I've selected has a dotted line previous to it, indicating the "Matching Skips Levels" feature is set to true. Look at the agreement. For the terminal node, the 'instance' encoded should match the instance encoded for "Phrase 3" (which is "Nominal 3" in the concise view). This just means the head term that is encoded at Phrase/Nominal 3 is the same as that at Terminal Node/Noun 3.

We do the same at the second object. This means that there can be any intervening structure whatever between "Nominal 3" and "Terminal Node+ 3", such as we see in Mt 3.11, but it will still match and return a hit. This search gives me 168 results in CascadiaSBLGNT; the previous (no agreement / skip levels) returned 73 hits.

Hope it helps explain a few things, particularly some of the power and flexibility that can be had using 'instance' agreement and "Matching Skips Levels".

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 25223
Forum MVP
Graham Criddle | Forum Activity | Replied: Fri, May 20 2011 2:12 PM

Rick

This is really useful and provides some additional insight (which I can always use!)

Thanks, Graham

Posts 19
Houston B Smith | Forum Activity | Replied: Fri, May 20 2011 5:16 PM

Thanks for your help.  It will take me some time to figure these searches out.

Page 2 of 2 (27 items) < Previous 1 2 | RSS