Syntactic search desiderata

Page 1 of 1 (5 items)
This post has 4 Replies | 1 Follower

Posts 5
Van Parunak | Forum Activity | Posted: Mon, Jul 16 2012 3:30 PM

The addition of syntactic annotations in both OT and NT and the ability to search them is a great step forward in Bible software. In using these capabilities, I've found myself wanting to do three things, but can't figure out how to construct the search. If these capabilities are present, would someone please tell me how to access them? If they are not, perhaps someone on the Logos team could consider them for the to-do list. I'll pose them in terms of the Andersen-Forbes OT, but they would be applicable to the Greek resources as well.

1. Sometimes I want to specify a node in a syntax graph, not in terms of its immediate labels, but in terms of lower level nodes (usually a segment) that it subsumes. For example, in searching for a noun phrase, I may not care whether it includes adjectives or an article, just that it contains a specific noun. E.g., in the first clause in the Bible, the direct object of ברא happens to be a union of two phrases, each including the object marker and an article as well as a noun. If I'm studying cooccurrence semantics of the verb, all I care about is cases where the object of ברא is (for example) ארץ, whatever its syntactic adornments. Symmetrically, I may want to find all cases where the predicator includes the verb ברא, whether or not  it has modifiers. The problem is complicated because the depth of the segment below the node that specifies its CIC function is not constant, so I can't just insert a bunch of dummy nodes. How can I specify a CIC in terms of segments that it contains, at whatever level?

2. In their recent book on phrase markers, AF emphasize the importance of tangled structures, those in which segments derived from a single parent are discontinuous in the text. I recently studied cases of paired subjects to a single predicator. I was interested in finding cases where predicator intervenes between the members of the paired subject (as in Isa 14:8, but not in 22:24). How can I construct a search that will return the first but not the second?

3. Sometimes it's important to find cases where a given CIC is NOT present in a clause. For example, we may want to find cases where a specific verb, usually transitive, appears without a direct object. How can I ask for such cases?

Thanks!

Posts 25105
Forum MVP
Dave Hooton | Forum Activity | Replied: Mon, Jul 16 2012 7:42 PM

Van Parunak:
1. Sometimes I want to specify a node in a syntax graph, not in terms of its immediate labels, but in terms of lower level nodes (usually a segment) that it subsumes. For example, in searching for a noun phrase, I may not care whether it includes adjectives or an article, just that it contains a specific noun. E.g., in the first clause in the Bible, the direct object of ברא happens to be a union of two phrases, each including the object marker and an article as well as a noun. If I'm studying cooccurrence semantics of the verb, all I care about is cases where the object of ברא is (for example) ארץ, whatever its syntactic adornments. Symmetrically, I may want to find all cases where the predicator includes the verb ברא, whether or not  it has modifiers. The problem is complicated because the depth of the segment below the node that specifies its CIC function is not constant, so I can't just insert a bunch of dummy nodes. How can I specify a CIC in terms of segments that it contains, at whatever level?

Right click the Segment and choose Matching skips Levels (often done automatically), which is indicated by a dotted line connector. Click the Segment to see the attributes you can select. Enter the lemma/lexeme as you would for a Morph Search.

With the Anything construct choose how many (eg. May or may not occur) syntax elements (eg. Clause IC's) may intervene; without it the Object must immediately follow the Predicator.

Hopefully you can see that Anything is needed to ignore/skip elements at the same level, whilst you can also specify matching to skip elements at different levels.

Dave
===

Windows 10 & Android 8

Posts 25105
Forum MVP
Dave Hooton | Forum Activity | Replied: Mon, Jul 16 2012 11:53 PM

Van Parunak:
2. In their recent book on phrase markers, AF emphasize the importance of tangled structures, those in which segments derived from a single parent are discontinuous in the text. I recently studied cases of paired subjects to a single predicator. I was interested in finding cases where predicator intervenes between the members of the paired subject (as in Isa 14:8, but not in 22:24). How can I construct a search that will return the first but not the second?

This is very difficult to specify precisely because the result segments do not follow manuscript order (the crossing lines in the Phrase Marker Analysis), and we have to use the Gap construct to get anywhere.

This will give you an idea:

It includes valid results like Is 14:8; 49:7 and excludes results like Is 22.24. See what happens when you remove Gap 1. To get a more precise result specifying a Predicator CIC is where the difficulty lies as it could be part of the Gap construct, or another Gap construct in another position!

Dave
===

Windows 10 & Android 8

Posts 25105
Forum MVP
Dave Hooton | Forum Activity | Replied: Tue, Jul 17 2012 1:09 AM

Van Parunak:
3. Sometimes it's important to find cases where a given CIC is NOT present in a clause. For example, we may want to find cases where a specific verb, usually transitive, appears without a direct object. How can I ask for such cases?

This should give you an idea

Here, we don't want a determiner (article) before the nominal phrase. Click the element/term box and choose "is not present" from the attribute that normally reads "is present".

Dave
===

Windows 10 & Android 8

Posts 5
Van Parunak | Forum Activity | Replied: Tue, Jul 17 2012 2:55 AM

Dave, your examples are most helpful. Thank you very much!

Page 1 of 1 (5 items) | RSS