Bug?: SGNT SyntacticForce omits many person names

Page 1 of 1 (12 items)
This post has 11 Replies | 1 Follower

Posts 27860
Forum MVP
Dave Hooton | Forum Activity | Posted: Fri, Nov 19 2021 4:15 PM

 <LN 93.1–93.388> NOT INTERSECTS {Section <SGNTSyntacticForce = prop. name>}  shows that the latter omits 439 person names in the ESV e.g. Jesus, John, Abraham!

The Louw-Nida subdomain is for Persons whereas the SyntacticForce term is for Persons and Places. Further, the subdomain excludes 64 occurrences of "Christ" that are not used as a person name vs. 7 for SGNT. "Lord" & "God" are excluded whereas SGNT includes 26 instances.

By comparison,  <LN 93.389-615> NOT INTERSECTS {Section <SGNTSyntacticForce = prop. name>} shows that the latter only omits 98 place names.

Dave
===

Windows 11 & Android 8

Posts 1466
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Mon, Nov 22 2021 8:18 AM

The Syntactic Force data (based on Lexham Syntactic Greek New Testament Data) was never intented to be comprehensive for each feature annotated (that would take a lifetime); instead, annotators focused on what they deemed to be the more significant annotations for the passage in question. So if a proper name was also the object of a preposition, it might be better annotated that way instead of as a personal name.

I'll file a case for some review of these particular items.

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 1361
LogosEmployee
Sean Boisen | Forum Activity | Replied: Mon, Nov 22 2021 9:47 AM

Dave, I'm curious: was there a particular outcome you were hoping to accomplish with this query? If so, I'd like to understand it and see if it's possible some other way (given the specific constraints on the SGNT data that Rick noted). 

Posts 1361
LogosEmployee
Sean Boisen | Forum Activity | Replied: Mon, Nov 22 2021 9:57 AM

Sean Boisen:

Dave, I'm curious: was there a particular outcome you were hoping to accomplish with this query? If so, I'd like to understand it and see if it's possible some other way (given the specific constraints on the SGNT data that Rick noted). 

Never mind: Andrew Batishko pointed me to the original thread at https://community.logos.com/forums/t/204002.aspx. I agree with your recommendation there to use LN codes. 

Posts 27860
Forum MVP
Dave Hooton | Forum Activity | Replied: Mon, Nov 22 2021 7:07 PM

Rick Brannan (Faithlife):

The Syntactic Force data (based on Lexham Syntactic Greek New Testament Data) was never intented to be comprehensive for each feature annotated (that would take a lifetime); instead, annotators focused on what they deemed to be the more significant annotations for the passage in question. So if a proper name was also the object of a preposition, it might be better annotated that way instead of as a personal name.

Sean Boisen:

Dave, I'm curious: was there a particular outcome you were hoping to accomplish with this query? If so, I'd like to understand it and see if it's possible some other way (given the specific constraints on the SGNT data that Rick noted). 

Sean,

I was expecting that a "proper name" (in the Context menu) will deliver that consistently, irrespective of the dataset. The SGNT Glossary states: "proper name: A form, often indeclinable, which expresses the formal name of a substantival element."  I couldn't find a Manual. So it is difficult to appreciate the limitations that Rick mentioned. One has to do some research for the Louw-Nida query but that was discovered by the user in the original thread, who just needed some Search assistance in determining the best query for his need.

Incidentally, The Bible Sense Lexicon manual states "A third perceived limitation may be the failure of the Bible Sense Lexicon to treat proper names. However, proper names are treated in the Biblical People and Biblical Places datasets within Logos Bible Software. This would have made treatment of proper names redundant." Not quite. Had it a <Sense Place Name>,  <Sense Person Name>, we could have a context sensitive tagging similar to Louw-Nida for the Old Testament (and the New).

Dave
===

Windows 11 & Android 8

Posts 1361
LogosEmployee
Sean Boisen | Forum Activity | Replied: Tue, Nov 23 2021 10:11 AM

Dave Hooton:

Rick Brannan (Faithlife):

The Syntactic Force data (based on Lexham Syntactic Greek New Testament Data) was never intented to be comprehensive for each feature annotated (that would take a lifetime); instead, annotators focused on what they deemed to be the more significant annotations for the passage in question. So if a proper name was also the object of a preposition, it might be better annotated that way instead of as a personal name.

Sean Boisen:

Dave, I'm curious: was there a particular outcome you were hoping to accomplish with this query? If so, I'd like to understand it and see if it's possible some other way (given the specific constraints on the SGNT data that Rick noted). 

Sean,

I was expecting that a "proper name" (in the Context menu) will deliver that consistently, irrespective of the dataset. The SGNT Glossary states: "proper name: A form, often indeclinable, which expresses the formal name of a substantival element."  I couldn't find a Manual. So it is difficult to appreciate the limitations that Rick mentioned. One has to do some research for the Louw-Nida query but that was discovered by the user in the original thread, who just needed some Search assistance in determining the best query for his need.

Incidentally, The Bible Sense Lexicon manual states "A third perceived limitation may be the failure of the Bible Sense Lexicon to treat proper names. However, proper names are treated in the Biblical People and Biblical Places datasets within Logos Bible Software. This would have made treatment of proper names redundant." Not quite. Had it a <Sense Place Name>,  <Sense Person Name>, we could have a context sensitive tagging similar to Louw-Nida for the Old Testament (and the New).

Dave, those are reasonable expectations. I'd still like to integrate our extensive name annotations into the Bible Sense Lexicon: omitting that was a pragmatic decision given the complexity of that work, not a desirable design. 

Posts 27860
Forum MVP
Dave Hooton | Forum Activity | Replied: Tue, Nov 23 2021 3:28 PM

Sean Boisen:
I'd still like to integrate our extensive name annotations into the Bible Sense Lexicon: omitting that was a pragmatic decision given the complexity of that work, not a desirable design. 

An alternative, often requested, would be <Person Any>, <Place Any> from the existing datasets.

Dave
===

Windows 11 & Android 8

Posts 1466
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Thu, Jan 20 2022 9:00 AM

The NT Syntactic Force dataset has been updated and the update should be available for licensed users.

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 27860
Forum MVP
Dave Hooton | Forum Activity | Replied: Sat, Jan 22 2022 3:19 AM

Hi Rick

The update fixes the person name discrepancy but still leaves a difference of 98 place names compared to  <LN 93.389-615>

Dave
===

Windows 11 & Android 8

Posts 5927
SineNomine | Forum Activity | Replied: Sat, Jan 22 2022 10:31 AM

Dave Hooton:
An alternative, often requested, would be <Person Any>, <Place Any> from the existing datasets.

Yes

Please use descriptive thread titles to attract helpful posts & save others time. Thanks!

Posts 1466
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Sat, Jan 22 2022 3:16 PM

Dave Hooton:
The update fixes the person name discrepancy but still leaves a difference of 98 place names compared to  <LN 93.389-615>

I read the "proper name" tag of the Lexham SGNT as applying to person names. But I can add the place names if you'd like (but they won't be distinguishable apart from combining with Louw-Nida).

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 27860
Forum MVP
Dave Hooton | Forum Activity | Replied: Sat, Jan 22 2022 3:37 PM

Rick Brannan (Faithlife):
I read the "proper name" tag of the Lexham SGNT as applying to person names. But I can add the place names if you'd like (but they won't be distinguishable apart from combining with Louw-Nida).

It already has at least 1023 place names but lacks 98 compared to Louw-Nida. Please add them

Dave
===

Windows 11 & Android 8

Page 1 of 1 (12 items) | RSS