Jump to content


Photo

syntactical databases


  • Please log in to reply
30 replies to this topic

#1 Dick Roberts

Dick Roberts

    Bronze

  • Active Members
  • PipPip
  • 65 posts
  • Gender:Male
  • Location:Vida, Oregon
  • Accordance Version:10.x

Posted 07 October 2008 - 12:58 AM

I was curious as to Accordance' future plans for implementing syntactical databases? Are there any current ones that could be licensed for Accordance?
Dick Roberts

#2 Helen Brown

Helen Brown

    Mithril

  • Admin
  • 8,500 posts
  • Gender:Female
  • Location:heart in Israel
  • Accordance Version:11.x

Posted 07 October 2008 - 05:16 AM

There are currently no syntactical databases available for Accordance. However, we are working on this issue, just want to get it right.
Helen Brown
OakTree Software

#3 Marco V. Fabbri

Marco V. Fabbri

    Silver

  • Accordance
  • 147 posts
  • Gender:Male
  • Location:Rome, Italy
  • Accordance Version:10.x

Posted 07 October 2008 - 08:35 AM

Dick,

what would you expect to find in a syntactical database? Maybe syntax within a sentence, like what the subject is, etc? Or do you rather think or larger units?

I have been thinking about this lately, and I wonder what would be more useful for Bible studies.

Marco
Marco Valerio Fabbri
P. UniversitÓ della S. Croce
Rome, Italy

#4 Clayton Willis

Clayton Willis

    Member

  • Members
  • Pip
  • 3 posts
  • Accordance Version:8.x

Posted 18 August 2009 - 11:04 AM

I would be interested in such a module. I think that the Accordance developers could come up with a great way (i.e. in a user-friendly and useful way) to display the information contained in the OpenText.org site.

In my thinking, Accordance's 'graphical' original language search interface would be an ideal way to set up the syntactical searches.

#5 RobM

RobM

    Gold

  • Active Members
  • PipPipPipPip
  • 230 posts
  • Gender:Male
  • Location:Chicago, IL
  • Interests:Hebrew, ANE languages and literature, Literary Theory
  • Accordance Version:11.x
  • Platforms:Mac OS X, iOS

Posted 18 August 2009 - 11:21 PM

Dick,

what would you expect to find in a syntactical database? Maybe syntax within a sentence, like what the subject is, etc? Or do you rather think or larger units?

I have been thinking about this lately, and I wonder what would be more useful for Bible studies.

Marco


Like Marco, I also question the usefulness of such a module. If it is limited to identifying how parts of speech are being used within a sentence (i.e., identifying that a given word is functioning as the subject, predicate, main verb, dependent clause, etc), then there could be a use for that.

However, if the intention is to go beyond this to identify syntactical uses of particular parts of speech (e.g., syntactical use of an aorist participle or genitive noun), then the module would be far too interpretive to provide any real help to the serious interpreter. After all, it is the job of the interpreter to identify these very kind of syntactical uses. A survey of translations and commentators reveals that there are very different interpretations of how certain parts of speech are functioning syntactically within a sentence.

So, like Marco asked, "what would you expect"? What would such a database look like?

#6 Clayton Willis

Clayton Willis

    Member

  • Members
  • Pip
  • 3 posts
  • Accordance Version:8.x

Posted 19 August 2009 - 12:42 AM

What I was thinking of would be very 'simple', but not necessarily just limited to identifying how the parts of speech are being used in a sentence. (i.e. I am NOT looking for a chart that tells me that this word is the subject, and this is the verb and this is the object etc...) What I envision would be access to the kind of information which would allow more precision when doing searches; for a simplistic example: we could search for a specific noun which is the subject of a specific verb, and find out what this particular syntactical relationship takes as it's object(s).

Clear as mud, yes? :huh:

#7 RobM

RobM

    Gold

  • Active Members
  • PipPipPipPip
  • 230 posts
  • Gender:Male
  • Location:Chicago, IL
  • Interests:Hebrew, ANE languages and literature, Literary Theory
  • Accordance Version:11.x
  • Platforms:Mac OS X, iOS

Posted 19 August 2009 - 10:40 AM

I think you can basically do that already. Simply do a Construct search for a particular noun in the nominative case (for greek) in the first column and the verb of your choosing (probably specifying indicative mood) in the second column have them occur within 10 or so words of each other, click the 'search both directions' box, click search. In your results window, narrow the search by 'clause'. that way you don't get hits where the words are clearly not grammatically tied together.

Just perform this search and read through the hits to see what kinds of objects the verb takes. Without having to translate everything it would not be too hard to identify whether or not it takes a direct object in the accusative or dative. Of course this takes a bit of reading/skimming.

Unless this is not what you are referring to.

Or are you thinking that such a database would remove the "reading/skimming" step I listed above?

Also, for this particular example, the standard lexicons already, to a significant degree, do that job. BDAG often includes what case a particular verb takes as its object (along with other grammatical, statistical, or etymological data). BDB will often tell you if a particular verb takes a particular preposition as its direct object marker (e.g., sometimes lamed [similar to aramaic] and beth are markers of the direct object for certain verbs)

Does this get close to what you are thinking of? Or, based on what I've written, can you (all) clarify a little more what you're thinking?

(note bene: this post reflects a development in my thoughts from beginning to end, i.e., I was "speaking my thoughts out loud")

#8 Helen Brown

Helen Brown

    Mithril

  • Admin
  • 8,500 posts
  • Gender:Female
  • Location:heart in Israel
  • Accordance Version:11.x

Posted 19 August 2009 - 11:17 AM

We are working towards providing syntactical information for the Greek and Hebrew texts that will be easy to both view and search. This is a massive project which will take some time to complete. Certainly the interpretation of the finer details will be more subjective than the broader classifications, but we aim to provide a system which will be understandable and useful to many users.
Helen Brown
OakTree Software

#9 Marco V. Fabbri

Marco V. Fabbri

    Silver

  • Accordance
  • 147 posts
  • Gender:Male
  • Location:Rome, Italy
  • Accordance Version:10.x

Posted 20 August 2009 - 10:57 AM

I may say that when I posted my earlier comment 10 months ago, I was not questioning whether some kind of syntactical tagging could be useful at all, but rather how it could best accomplished.
I was fully aware that some subjective judgment would be involved, but that wasn't driving me away.

Let us say that the grammatical tagging is often considered as more objective, as opposed to the more subjective syntactical tagging. But then, it was apparent to me that in some cases the grammatical tagging depends on the interpreter's view on the syntax.

This is because the grammatical tagging requires the person who prepares the tags to make some decisions. This happens as in Greek (this applies, of course, also to other languages) there are different forms that are homographs.

The easiest example are neutral articles or pronouns. It is impossible to say whether they are accusative or nominative without looking at the context. And the context which is relevant is the syntactical context.

I can see two ways of solving this problem.

One would be to indicate all the possible cases that can be represented by a given inflected form. This would lead to a database that indicates nominative, accusative or even vocative as possible cases of a given inflected form. This is not the way GNT-T, or LXX or VULG-T are tagged.

The second way is to make choices. This requires an interpreter who makes choices, and the choices are based on the syntactical context. In some cases at least, the choice is so hard, that it is best to indicate alternate parsings. This does not mean that it would be better to give alternate parsings for every several occurrence of the Greek neuter article TO or TA. This is the way that actual Accordance modules as GNT-T, LXX or VULG-T were prepared.

If so, I thought that it would be extremely interesting to try to prepare a syntax module, and to do that in such a way that would reduce alternate interpretations to a minimum. This is where I was when I wrote ten months ago.

I see that now interest for the issue is rekindled, and I anticipate an interesting discussion.

Edited by Helen Brown, 20 August 2009 - 11:41 AM.

Marco Valerio Fabbri
P. UniversitÓ della S. Croce
Rome, Italy

#10 Marco V. Fabbri

Marco V. Fabbri

    Silver

  • Accordance
  • 147 posts
  • Gender:Male
  • Location:Rome, Italy
  • Accordance Version:10.x

Posted 21 August 2009 - 05:34 AM

I think you can basically do that already. Simply do a Construct search for a particular noun in the nominative case (for greek) in the first column and the verb of your choosing (probably specifying indicative mood) in the second column have them occur within 10 or so words of each other, click the 'search both directions' box, click search. In your results window, narrow the search by 'clause'. that way you don't get hits where the words are clearly not grammatically tied together.

(snip)


Rob:

you are right, this is a good way to do something that gets close. If you don't mind, I would consider the same issue within my previous line of thought, that tended to draw attention to the elements that should be included in a module that tags syntax.

You suggested to narrow the search by clause, so as not to get hits where the words are not grammatically connected. Now, my question is: how is a clause determined? That would influence greatly the result of your search. How does the search engine know about clause boundaries?

As I see it, it could know in two ways. The first way would be an automated way: that would mean using some marker that is already in the text as a clause boundary. What could these markers be? I suggest that they consist of punctuation: commas, semicolons, etc. This would make sense: fullstops would be separators for sentences, and commas or similar would be separators for clauses.

However, this would hardly be a reliable tool to make sure that the words in your search are, or are not, grammatically connected. A clause could easily include a comma, and so extend further than the comma. Or a sentence could not include any punctuation mark, and so there would be no way for the search engine to divide it into clauses.

If we are not to give up and renounce the benefits of a search restricted to clauses, then we will need a module that includes tags that mark the beginning and the end of clauses. This can be tricky, as there clauses included within longer clauses, and even clauses that split into parts that are not necessarily contiguous.

This would require the tagger to make decisions on every several sentence of the text. Decisions of this kind are based on syntax, not on morphology alone. Of course, many of these decisions could be challenged and discussed. However, as long as the system is consistent, the result would greatly improve the significance of a search like that described.

From this I would draw the conclusion that a desirable feature of a syntactical module would be that of systematically defining the boundaries of clauses.

Edited by Marco, 21 August 2009 - 05:37 AM.

Marco Valerio Fabbri
P. UniversitÓ della S. Croce
Rome, Italy

#11 RobM

RobM

    Gold

  • Active Members
  • PipPipPipPip
  • 230 posts
  • Gender:Male
  • Location:Chicago, IL
  • Interests:Hebrew, ANE languages and literature, Literary Theory
  • Accordance Version:11.x
  • Platforms:Mac OS X, iOS

Posted 21 August 2009 - 12:35 PM

Marco,
I'm sorry I misrepresented you in my previous post.

Your two posts, it seems to me, only accentuate my first point that a syntax database would require a lot of interpretive decisions (i.e., subjectivity). My suggestion for how "you can basically do that already", I grant you was limited (and your point about the clause search is perfectly valid especially with passages like Eph 1 where almost the whole chapter is one sentence where phrases in later verses are grammatically/syntactically tied to the 3rd, 4th or 5th [etc] verse prior).

My question is, due to the subjective nature of such a database, is it really worth having (and spending valuable Accordance developer time on)? Again, your comments about how the grammatical tagging is at times subjective makes me doubt even more the reliability of a syntax database. Am I off base here?

Now, I recognize that such a database could/should be perceived in a way similar to how a commentary is perceived: a resource filled with informed interpretations. Perhaps such a database should be called a commentary (for example, "New Testament Syntactical Commentary") rather than a database ("New Testament Syntactical Database"). I guess I wonder about how such a database will be used. Will it be used competently as an interpretive resource or used incompetently as an answer book for the lazy student or (dare I say) pastor?

#12 Marco V. Fabbri

Marco V. Fabbri

    Silver

  • Accordance
  • 147 posts
  • Gender:Male
  • Location:Rome, Italy
  • Accordance Version:10.x

Posted 21 August 2009 - 12:46 PM

Marco,
I'm sorry I misrepresented you in my previous post.

(snip)


Rob:

there was nothing wrong with your comment.

Your question about how subjective a syntactical database is is a good one. I tried to show that also morphological databases involve subjective decisions on part of the developer. Apart from mistakes that can always creep in, but can also be corrected in later updates, there are some choices that could be made one way or the other. This is true of dictionaries, grammars, translations. It is also true of a syntactical database.

You also point to the reasonable answer: users should be aware that the syntactical database contains informed interpretation. The learner can use it as a guide to the study of syntax, just as if he had the help of a living teacher. And the scholar can use the database to gather quickly information that he considers relevant. Then he can disagree on any of the results. I hope that he would disagree in some of them, not in most of them. If so, the database would be useful even for him.

I think that this discussion has already brought out some characteristics that are desirable in a syntactical database: it should define the boundaries of clauses, it should indicate subject, predicates, and few other syntactical functions, it should be possible to make searches in both the main search window and the construct window.

What else am I missing?

Edited by Marco, 22 August 2009 - 10:21 AM.

Marco Valerio Fabbri
P. UniversitÓ della S. Croce
Rome, Italy

#13 RobM

RobM

    Gold

  • Active Members
  • PipPipPipPip
  • 230 posts
  • Gender:Male
  • Location:Chicago, IL
  • Interests:Hebrew, ANE languages and literature, Literary Theory
  • Accordance Version:11.x
  • Platforms:Mac OS X, iOS

Posted 24 August 2009 - 04:34 PM

And the scholar can use the database to gather quickly information that he considers relevant. Then he can disagree on any of the results. I hope that he would disagree in some of them, not in most of them. If so, the database would be useful even for him.


Marco,
I guess I wonder if it would be possible for even this to be so. For example: I think that Wallace's Exegetical Syntax is currently the closest thing to a syntactical database (though it doesn't do some or most of what you are talking about for the desired database). And Wallace notes a great number of places where the grammar is either ambiguous or more research still needs to be done or there is disagreement down theological lines (where both sides still think the other is wrong).

I think that this discussion has already brought out some characteristics that are desirable in a syntactical database: it should define the boundaries of clauses, it should indicate subject, predicates, and few other syntactical functions, it should be possible to make searches in both the main search window and the construct window.

What else am I missing?


It may also be useful for the database to include cross references to standard grammars like BDF and Wallace's Exegetical Syntax. In fact, these may be a good starting point in gathering syntactical constructions that have already been figured out and are accepted as normative.

What do you think?

#14 Marco V. Fabbri

Marco V. Fabbri

    Silver

  • Accordance
  • 147 posts
  • Gender:Male
  • Location:Rome, Italy
  • Accordance Version:10.x

Posted 28 August 2009 - 12:45 PM

Rob:

I have gone through Wallace before answering. The good thing is that it is already available as a module within Accordance. As everything is connected with hyperlinks, you can easily move from GNT-T to Wallace and back. So, your idea proved a useful tip for me.

As for ambiguities: there are certainly many of them, but they are not so many as to make a syntactical database unreliable. Rather, a user should be aware that such a module implies choices, and that a certain amount of them can be challenged. In other words, results in a search can't be 100% exact. Even so, I find that a high percentage of choices that are reliable enough.

For instance, Wallace discusses how predicate nominatives can be distinguished from subject nominatives. The treatment is sound, and provides a good starting point for accurate tagging.

As for the research needs to be done: I tend to think that preparing or using a syntactical module may prove a good way to do that research.
  • RobM likes this
Marco Valerio Fabbri
P. UniversitÓ della S. Croce
Rome, Italy

#15 Daniel R

Daniel R

    Member

  • Active Members
  • Pip
  • 48 posts
  • Gender:Male
  • Accordance Version:11.x

Posted 26 September 2009 - 05:39 AM

We are more interested in the ability to build our own databases and use them in Accordance. Rather than simply get licenses for existing syntactic databases (which would still be nice), we wish to make our own morphological (Westminster for BHS isn't perfect), syntactic, and semantic databases.

Edited by danielandtonya, 26 September 2009 - 05:39 AM.


#16 Helen Brown

Helen Brown

    Mithril

  • Admin
  • 8,500 posts
  • Gender:Female
  • Location:heart in Israel
  • Accordance Version:11.x

Posted 26 September 2009 - 12:19 PM

Sorry, this is not part of our plans. I think that it would be extremely difficult, and create a host of support issues.
Helen Brown
OakTree Software

#17 Daniel R

Daniel R

    Member

  • Active Members
  • Pip
  • 48 posts
  • Gender:Male
  • Accordance Version:11.x

Posted 26 September 2009 - 01:49 PM

Sorry, this is not part of our plans. I think that it would be extremely difficult, and create a host of support issues.



Can you, or anyone, think of a Mac program that would do the job? As of now, we have to use a program in Windows XP to make our own syntactic database and lexicon.

Edited by danielandtonya, 26 September 2009 - 01:55 PM.


#18 Marco V. Fabbri

Marco V. Fabbri

    Silver

  • Accordance
  • 147 posts
  • Gender:Male
  • Location:Rome, Italy
  • Accordance Version:10.x

Posted 27 September 2009 - 08:02 AM

Can you, or anyone, think of a Mac program that would do the job? As of now, we have to use a program in Windows XP to make our own syntactic database and lexicon.


If you want to take your notes, Accordance offers a layout. You need to select a portion of text, and then click the "syntax" button in the amplify palette. You will be presented with a table, were everything is already compiled except the syntactical analysis. You can easily enter your own assessment of the syntax.

However, this is not a module, but rather the output of a grammatically tagged module, arranged so that you can add to the information already provided. I can't be searched within Accordance.

Putting together a database that can be searched with the Accordance powerful engine would require a group of scholars to agree first about a uniform understanding of syntax, and then also about a uniform format for collecting the information. This is way, I think it would be way too difficult to have everyone to choose his own rules.

What, then, if there were some important difference in the way syntax is understood and represented? I would say that having a module that reflects one way of understanding syntax would open the way to more modules that reflect a different understanding. It would be a problem similar to the problem of choosing which text should be chosen for a tagged module of say, the Hebrew Bible or the NT. Is it better to have a critical edition of the NT? Or the majority text? Or a single manuscript? Accordance started with the Greek New Testament, and then added the GNT by Wetscott-Hort, the GNT by Tischendorf, the Codex Vaticanus, the Codex Bezae and so on. Now we can choose according to our need. The was a time when we had nothing, then a time when we had only one text. Something similar might happen with syntax modules.
Marco Valerio Fabbri
P. UniversitÓ della S. Croce
Rome, Italy

#19 RobM

RobM

    Gold

  • Active Members
  • PipPipPipPip
  • 230 posts
  • Gender:Male
  • Location:Chicago, IL
  • Interests:Hebrew, ANE languages and literature, Literary Theory
  • Accordance Version:11.x
  • Platforms:Mac OS X, iOS

Posted 27 September 2009 - 02:23 PM

Putting together a database that can be searched with the Accordance powerful engine would require a group of scholars to agree first about a uniform understanding of syntax, and then also about a uniform format for collecting the information. This is way, I think it would be way too difficult to have everyone to choose his own rules.

What, then, if there were some important difference in the way syntax is understood and represented? I would say that having a module that reflects one way of understanding syntax would open the way to more modules that reflect a different understanding. It would be a problem similar to the problem of choosing which text should be chosen for a tagged module of say, the Hebrew Bible or the NT. Is it better to have a critical edition of the NT? Or the majority text? Or a single manuscript? Accordance started with the Greek New Testament, and then added the GNT by Wetscott-Hort, the GNT by Tischendorf, the Codex Vaticanus, the Codex Bezae and so on. Now we can choose according to our need. The was a time when we had nothing, then a time when we had only one text. Something similar might happen with syntax modules.


Marco,

That's an interesting observation and prediction. From our conversation before I feel won over to supporting this kind of module. With this comment, I am more curious to see what the possibilities may be....

Off the top of my head, I can think a few possible directions: a reformed syntax module, a dispensational syntax module, an arminian syntax module, a calvinist syntax module, a feminist syntax module, a conservative module, a liberal module. (after writing that out, I begin to feel again my skepticism about this kind of module).

However, I think you are right in one or two of your other posts where you mention that there are many constructions that are objective and regular that can be documented in a syntax module that all can easily agree upon. For other constructions, I think various options would need to be included. What I mean is, when perusing the results of a particular search in the syntax module there should be notification of passages where the syntax is disputed (in other words, the search performed is merely one option for a given passage while there are people/traditions that see the syntax differently). And I think there should be some way of viewing those different options in situ: by opening a new pane or by hovering over a link and viewing it in the ID palette.

#20 Daniel R

Daniel R

    Member

  • Active Members
  • Pip
  • 48 posts
  • Gender:Male
  • Accordance Version:11.x

Posted 28 September 2009 - 01:57 AM

If you want to take your notes, Accordance offers a layout. You need to select a portion of text, and then click the "syntax" button in the amplify palette. You will be presented with a table, were everything is already compiled except the syntactical analysis. You can easily enter your own assessment of the syntax.

However, this is not a module, but rather the output of a grammatically tagged module, arranged so that you can add to the information already provided. I can't be searched within Accordance.

Putting together a database that can be searched with the Accordance powerful engine would require a group of scholars to agree first about a uniform understanding of syntax, and then also about a uniform format for collecting the information. This is way, I think it would be way too difficult to have everyone to choose his own rules.

What, then, if there were some important difference in the way syntax is understood and represented? I would say that having a module that reflects one way of understanding syntax would open the way to more modules that reflect a different understanding. It would be a problem similar to the problem of choosing which text should be chosen for a tagged module of say, the Hebrew Bible or the NT. Is it better to have a critical edition of the NT? Or the majority text? Or a single manuscript? Accordance started with the Greek New Testament, and then added the GNT by Wetscott-Hort, the GNT by Tischendorf, the Codex Vaticanus, the Codex Bezae and so on. Now we can choose according to our need. The was a time when we had nothing, then a time when we had only one text. Something similar might happen with syntax modules.


So that's a no. We are well familiar with Accordance's features and the syntax feature does not do the job we're talking about. We completely disagree with the thought that scholars would have to agree on the syntax to make a database. We are interested in developing our own thoughts, not just using others.

Again, can anyone name a Mac program that will make language databases (and not just syntax)?




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users