Abstract

A manuscript comparison has two distinct responsibilities: preserve the witness being displayed, and help readers understand it. Urevangelium treats the recorded ancient-language wording and sequence as fixed inputs to the reading controls. English is a separate, attributed layer. A reader can place it beside the corresponding source words or read it independently. Where a complete scholarly translation is registered, its published sequence is used. Where only sourced glosses are available, a small, disclosed rule set may rearrange those glosses. It supplies no new English wording and makes no changes to the ancient text.

1. Different languages arrange meaning differently

Readers are right to care about faithful English. Faithfulness, however, cannot be measured simply by whether English words occupy the same positions as the words of another language. Greek inflections carry grammatical information that English often expresses through position. Greek word order also contributes emphasis and discourse structure; it is not arbitrary. Goodell’s grammar discusses both the differences from English and restrictions within Greek. See Goodell, §§674–676, and the Open University’s introduction to word order.

This is why Urevangelium preserves the ancient sequence even when an English reading uses a different arrangement. The distinction concerns what the interface displays. It does not imply that English order is meaningless, that all arrangements are equally accurate, or that one manuscript settles every textual question.

2. Three separate layers

Ancient source

The selected witness or edition supplies the source words and their recorded order. Neither English mode may insert, delete, substitute, duplicate, or move them. Selecting another witness, such as Sinaiticus instead of Vaticanus, is a separate operation.

English wording

A registered scholarly translation or source-attributed gloss supplies the English. A gloss is a word or phrase aid, not necessarily a complete sentence translation. The STEPBible TAGNT documentation, for example, describes contextual translations attached to Greek word records. A dictionary sense, a contextual gloss, and a published verse are distinct kinds of evidence.

English arrangement

The interface decides where existing English occurrences appear. This layer can move references to English words or gloss groups. It cannot author substitute wording. A repeated name remains a distinct occurrence each time it appears.

3. An example: Matthew 1:7

The project’s Clementine Latin begins Salomon autem genuit Roboam. The Douay–Rheims 1899 translation begins “And Solomon begot Roboam.” The connective occupies a different position. The English words can be displayed in either of these two arrangements:

The Latin remains in the first sequence in both modes. The English follows the chosen display rule. Hovering or focusing an assigned English occurrence still identifies its source counterpart. This example is an illustration of an existing occurrence mapping, not permission to infer every mapping from English spelling. Inspect Matthew 1:7.

4. What the two controls do

Follow source keeps English beside its assigned ancient-language word or word group. An English phrase may correspond to one source word. Unassigned published English remains visible as awaiting alignment; the program does not force a match merely to fill a space.

Translation order displays the English independently. Vulgate, Peshitta, and the optional Sinaiticus column use their registered Douay–Rheims, Murdock, and Anderson translation units when available. Published wording, punctuation, and sequence are retained. If verse numbering differs, the published verse and its documented boundary mapping remain distinct.

For columns without a registered complete translation unit, the label System-arranged glosses identifies the alternative. It is a reading aid assembled from existing attributed glosses, not a quotation from an otherwise unregistered English Bible. The same unobtrusive controls appear over every English subcolumn. Bezae’s two halves contain Greek and Latin manuscript text, so neither is an English subcolumn and neither is reordered by these controls.

5. What the automation actually does

The current automatic arrangement is deliberately limited. It makes a separate list of references to the existing gloss groups. A recognized Greek or Coptic postpositive connector can move before the immediately preceding group when its existing English gloss is an isolated connective such as “and,” “then,” or “for.” The program does not change that gloss to a different connective.

The implemented source-form list is Greek δε, γαρ, ουν and Coptic ⲇⲉ, ⲅⲁⲣ, ⲟⲩⲛ. Recognition ignores Greek accent marks for this comparison only; the displayed source forms remain untouched. The move is withheld across terminal punctuation, registered phrase spans, missing English, another recognized connector, or a proposed papyrus reading. All other groups retain their existing relative order. English-only expansion groups remain separately identifiable.

These are project display rules. They are not a general parser of Greek, Coptic, Latin, or Syriac and do not guarantee a fluent English sentence. Goodell’s discussion of postpositive position motivates separating ancient and English order; it does not independently validate this algorithm or its Coptic application. Complex clauses and the accuracy of individual gloss choices still require appropriate linguistic review.

No language model is called to generate English. The algorithm cannot add pronouns, replace names, choose synonyms, paraphrase a clause, complete a damaged manuscript, or import wording from a neighboring tradition. Word groups, source attribution, and proposed-reading marks travel together.

6. Missing evidence remains visible

The automatic reading excludes project-generated English, experimental wording, automatic name annotations, and alignment placeholders. Where a source-attributed gloss is unavailable, an em dash and an explanatory note remain. This avoids presenting a program’s own wording as a scholarly translation. Existing annotations in Follow source retain their existing labels; the new ordering feature does not certify them.

A scholarly gloss attached to a proposed reading in a damaged papyrus keeps its dagger. It does not become evidence that missing Greek letters survive. Physical manuscript loss, textual omission, an alignment gap, and data not yet populated remain separate states. English display choices cannot resolve those differences.

7. Checks and limits of the claim

Verification compares source occurrence identifiers, text, and order before and after switching. The registered translation path checks the published English bank and its occurrence accounting. The gloss path checks that eligible English groups are preserved exactly once with unchanged wording and attribution, that excluded annotations do not enter the reading, and that source arrays are not mutated. Keyboard and hover checks ensure English still points to its recorded source occurrence after it moves.

These checks establish specific software properties. They do not establish that every source transcription is correct, every English gloss is the best contextual interpretation, or every alignment has received independent scholarly review. A faithful interface must keep those questions open wherever the evidence remains open. For the governing sources and certification status, consult the tradition certification systems.