Switch branches/tags
Nothing to show
Find file History
Latest commit 707d461 Jul 13, 2018

README.md

Nestle1904 Lowfat: Greek New Testament Syntax Trees

https://github.com/biblicalhumanities/greek-new-testament/syntax-trees/nestle1904

This repository contains syntax trees in a format useful for programmers and data analysis. Users who simply want to look up the structure of a sentence can see this data with a very simple user interface here:

These syntax trees were created to be easier to query and to display using standard Web technologies. They are called "lowfat" because they are simpler and have fewer nodes than the equivalent GBI trees: about 1/2 as many elements, and about 1/3 as many attributes.

They are based on the 1904 edition of Eberhard Nestle's Greek New Testament, often referred to as the Nestle 1904 or British Foreign Bible Society 1904. If you need the original paragraph markings, quotations, and contextualized glosses for this text, please see the following repository:

Please report bugs by creating issues on this github directory. You can also contact us at:

For other high quality resources suitable to scholarly study of Greek and Hebrew, see the Biblical Humanities Dashboard.

Copyright

Syntax diagram markup copyright 2014-2017 by Jonathan Robie and Micheal Palmer, made available under under a [CC-BY-SA License] (http://creativecommons.org/licenses/by-sa/3.0/).
Created by a transformation from Syntax diagram markup copyright 2014-2015 by the Global Bible Initiative, made available under under a [CC-BY-SA License] (http://creativecommons.org/licenses/by-sa/3.0/).

Morphological parsing and lemmatization by Dr. Ulrik Sandborg-Petersen of Emergence Consult and Aalborg University, Denmark, and released into the public domain. Available at https://github.com/biblicalhumanities/Nestle1904.

Greek text transcribed by Diego Renato dos Santos and released to the public domain, available at https://sites.google.com/site/nestle1904/.

These syntax trees were automatically generated, then carefully corrected by hand. Please report any errors you encounter.

Markup scheme

This is an example of the current markup scheme:

  <milestone unit="verse" n="John.11.35">John.11.35</milestone>
   <wg class="sentence">
      <wg class="cl">
         <wg role="v" class="vp" head="true">
            <w class="verb"
               osisId="John.11.35!1"
               lemma="δακρύω"
               normalized="ἐδάκρυσεν"
               strong="1145"
               number="singular"
               person="third"
               tense="aorist"
               voice="active"
               mood="indicative"
               head="true">ἐδάκρυσεν</w>
         </wg>
         <wg role="s" class="np" articular="true" det="true">
            <w class="det"
               osisId="John.11.35!2"
               lemma=""
               normalized=""
               strong="3588"
               number="singular"
               gender="masculine"
               case="nominative">ὁ</w>
            <wg class="np" head="true">
               <w class="noun"
                  type="proper"
                  osisId="John.11.35!3"
                  lemma="Ἰησοῦς"
                  normalized="Ἰησοῦς"
                  strong="2424"
                  number="singular"
                  gender="masculine"
                  case="nominative"
                  head="true">Ἰησοῦς</w>
               <pu>.</pu>
            </wg>
         </wg>
      </wg>
   </wg>

Here is an overview of this format:

  • <sentence> is, obviously, a sentence. Sentences are determined based purely on punctuation used in the original text.
  • <cite> is a citation that identifies the text in a sentence.
  • <wg> stands for "word group", and is used to identify a group of words.
  • <w> stands for "word", and identifies a single word.

A <wg> can have the following attributes:

  • nodeId: an identifier for the <wg> node
  • class: the class of the word group. Permitted values: np cl pp vp adjp advp nump adv conj
  • role: the clause-level role of the world group. Permitted values: s v vc o p io o2 adv

A <w> can have the following attributes:

  • morphId: identifier for the word occurrence
  • class: the class of the word. Permitted values: noun verb det conj pron prep adj adv ptcl num int
  • role: the clause-level role of the word. Permitted values: s v vc o io o2 p adv
  • head: 'true' if the word is the head of the phrase.
  • discontinuous: 'true' if the word is discontinuous with respect to sentence order due to reordering in the syntax tree
  • lemma: the lemma of the word
  • person: first, second, or third
  • number: singular or plural
  • gender: masculine, feminine, or neuter
  • case: nominative, genitive, dative, accusative, or vocative
  • tense: aorist, present, imperfect, future, perfect, or pluperfect
  • voice: active, passive, middle, or middlepassive
  • mood: indicative, imperative, subjunctive, optative, participle, or infinitive

Python Tutorial

Here is an introductory tutorial showing how to use the Lowfat syntax trees using Python and XPath in a Jupyter Notebook: