Comparing NT Syntax Searches

Dave Hooton

I thought to compare OpenText and the new Lexham Syntax Graphs with a simplified Genitive Absolute search as below

 

The results are Lexham 123, OpenText 91.

Lexham has 4 additional verses (Mk 6:2; 6:22; Rom 8:11; 9:16) and OpenText has 1 additional verse at Mk 10:46. These are due to different analysis.

The additional results in Lexham are due to additional levels of the Subject structure which can be restricted to the "Last" in OpenText eg. Acts 15:2. How could I do this in the Lexham structure?

Rick Brannan (Logos)

The additional results in Lexham are due to additional levels of the Subject structure which can be restricted to the "Last" in OpenText eg. Acts 15:2. How could I do this in the Lexham structure?

The Lexham Syntax Graphs bubble the lemma/morph/LN data for the head term up the graph, so the actual head term of each node has data available on each node. So technically, if you just want to find genitive absolutes (but not necessarily highlight the words that make them up) you don't even need the word node, you can specify case, mood, part-of-speech, lemma or LN data (article, subdomain, domain) at the node (here the "clause function") -- and not worry about the word object.

This may not have fully made it to the version released with beta 4, and if it did it has errors in the head term and morph data. These have been corrected and I'm hoping the updated resource makes it into beta 5. But I just ran a search on Lexham Syntax Graphs (the corrected resource) and got 93 results. You'll likely get different results with your beta 4 version of the syntax database resource. A graphic of the query is below (first graphic). Note that I'm not saying this finds all genitive absolutes, I'm just trying to reconstruct what it seems you're doing.

Further note: If you do want to highlight words, you could add a word object to each clause function that skips intervening levels. In future versions of the database, there will likely be a "word identifier" or something so you can specify agreement in "word identifier" between the node (here "clause function") and the word, to ensure you're highlighting the same thing. In the mean time, you can cheat by agreeing in LN Article because these are largely unduplicated in this type of context (this is the second graphic below; both queries have exactly the same hits, the second just highlights words). Also, I know there is inconsistent terminology in this dialog ("Head Text & Lemma" but "Term Morphology"); this will be clarified at a later date as well (likely not beta 5).

[second query, this one highlights the necessary words]