10 replies to this topic

[rpavich]

Coming from the "other guys" bible software I'm used to doing certain things a certain way, and I realize that I'll have to relearn a LOT of things.

 

One thing that I'd like to be able to do is construct a syntax search. I checked out the help file but it's a lot easier to watch someone do it and talk about it.

 

Would you do a podcast on doing simple constructs in english and greek/hebrew?

 

One example that I had in mind was the answer to this question: Find any clause where God is the subject but not necessarily referred to by the word "God"

 

Or another example; Find all the places where God is the subject and displays an emotion. (again, not a simple "God" word search.

 

 

Does this make sense?

[Timothy Jenney]

Hi!

 

There are already a number of podcasts on construct searches: http://www.accordanc...uctsearches.m4v, http://www.accordanc...kconstructs.m4v, and http://www.accordanc...wconstructs.m4v, as well as one on syntax: http://www.accordanc...iontosyntax.m4v.

 

Unfortunately, a construct search, even one using using syntax, will not find the clauses you want. I see that someone else has mentioned this to you in another thread. He's right. The syntactical database we have does not identify the antecedent of pronouns, even when they are found in the same sentence. It does not address the logical (or grammatical) relationship between sentences at all.

 

Sorry.

[דָנִיאֶל]

I had thought that the purpose of the Antecedent element in the construct was to handle things such as this. It is possible to construct searches where one can find something tagged as an antecedent which is the lex θεος for example. I haven't tried to fully work up the example yet and it would require being able to handle the null cases. So is the real issue not that antecedents are not tagged but the thing that they are the antecedent of is not indicated ? If so that would be a nice enhancement. In the meantime it would seem you could find sentences containing pronouns and antecedents which are God. An alternative would appear to be a topic search.

 

Thx

D

[Robert Holmstedt]

The syntax tagging only tracks antecedents for syntactic purposes within a single clause domain, i.e., resumption for dislocation (aka casus pendens) or the head of a relative clause. Tracking broader issues of antecedence would be roughly the same thing as tracking participant reference, which is both basically a discourse concern (i.e., the fact that a pronoun in one clause refers back to the noun God in a previous clause) and horribly, nightmarishly complicated (and so would not only bloat the syntax module but would have pushed everything off by at least 5 more  years). And anyway, one can usually find the reasonably accurate antecedent relationship by scanning a discourse selection the old-fashioned way. 

[Robert Holmstedt]

Since this started with reference to the Andersen-Forbes database in Logos' package, it is worth reminding users that what we've produced in 5 years took Frank Andersen and Dean Forbes over 30 years. Moreover, since I was not that interested in tracking things like emotions or animal types and such, I decided to push our  syntactic tagging into more texts (i.e., DSS and epigraphy) rather than towards more and more (and more and more, endlessly) semantic/discourse/lexical categories.

[rpavich]

Thanks everyone for replying and I certainly meant no disrespect to any dev or programmer by my question.

[Robert Holmstedt]

None taken. It's simply important that expectations are set accordingly. The A-F database is much, much more than syntax (and took much longer).

[דָנִיאֶל]

Thanx for the clarifications here Robert. It makes better sense of the comment in the syntax guide PDF where it says antecedents are sometimes noted.

 

Regarding discourse tagging for such things as referent relationships, I have been wondering if there is much interest in producing such a thing in Accordance. I would be interested in any thoughts you might have on this.

 

Thx

D

[Robert Holmstedt]

It's a dicey project. First, it would require someone with an immense amount of time. Second, it would have to be yet another layer of tagging on top of the syntax (which is layered on top of the morph data). It's hard to describe the challenges with this unless you see what our underlying text looks like. Very briefly, to connect multiple antecedents, sometimes across rather large stretches of discourse, makes an incredibly complicated text to deal with. I can see some uses of such a database, but the thought of it being done makes me sigh very deeply.

[דָנִיאֶל]

Thanx for your thoughts. I appreciate all these layer-the-text-with-another-db projects are time-consuming and fraught with difficulties of one kind or another. The outcome is very useful though.

 

Thx

D

[rpavich]

Thanks for explaining; as a layman one project seems the same as any other to me; I'm not privy to what it takes to do any of this.