This document outlines the major challenges, limitations, and editorial decisions involved in constructing the Biblical alignment section of the Parallelium dataset.
Aligning Biblical texts at scale involves confronting the complex transmission history that spans religious traditions, languages, and manuscript cultures. This requires addressing significant structural and philological variation. These challenges include differences in book organization, verse numbering, canonical content, and translation lineage.
The alignment strategies described here aim to support computational training and evaluation, rather than scholarly critical editions or theological study. As such, certain compromises and simplifications were made to enable consistency, transparency, and reusability.
The following sections provide examples of structural divergence across traditions, justification for exclusion or realignment of specific books or verses, and a clear statement of the dataset's intended use and limitations.
-
While modern Bibles are widely available online, medieval versions are rare and often exist only in print or in inaccessible formats.
-
Encoding inconsistencies and varying editorial conventions across sources require substantial normalization and preprocessing.
-
The texts in this dataset originate from multiple religious and textual traditions, requiring careful textual literacy to align responsibly.
-
Canonical order and verse mapping vary significantly between traditions. Books may differ not only in name, but also in structure—being combined, split, reordered, expanded, or labeled differently across canons. These structural divergences directly influence alignment choices.
- In the Latin Vulgate, Ezra and Nehemiah appear as 1 Esdras and 2 Esdras.
- In the Septuagint, 1 Esdras (Esdras A) is a distinct book that overlaps partially with Ezra and includes unique material (e.g., the “Three Bodyguards” episode).
- 2 Esdras (Esdras B) in the Septuagint aligns more closely with the Hebrew and Latin narrative but uses a different naming convention.
- The book of Daniel includes supplementary material such as Susanna, Bel and the Dragon, the Prayer of Azariah, and the Song of the Three Young Men.
- In the Septuagint, the Prayer and the Song are embedded in Daniel 3; in the Vulgate, they are often marked as separate insertions with distinct headings.
- Some traditions split or combine verses differently. For example, the Epistle of Jeremiah is presented as Baruch 6 in the Vulgate.
- The Psalms are notoriously variable in numbering across Hebrew, Greek, and Latin versions, complicating verse-level mapping.
- Psalmus 151 appears in the Septuagint and some Vulgate manuscripts, but it lacks a standardized placement and is often excluded from canonical Psalters.
Even when books are nominally shared, deep structural differences often prevent straightforward alignment. In such cases, editorial intervention—such as verse-splitting, segment reordering, or exclusion—may be required to ensure integrity and consistency.
It does not preserve canonical verse numbering, nor does it claim textual completeness or editorial neutrality. As such, it is not suitable for:
- Scholarly editions or critical apparatus
- Canonical or liturgical citation
- Fine-grained textual-critical analysis
➡️ Return to Dataset Overview