[ronaldtse]

ALA-LC batch 2 — Arabic-script / Hebrew-script systems (11 maps)

Second batch of the remaining-encodings campaign's ALA-LC work order. Each map transcribes its own source document (read visually, codepoint-verified against the text layer) and carries embedded tests taken from that document's tables and rule examples. Nothing pre-existing is touched; the change only adds files.

Systems encoded

Map	Source document	Tests
alalc-kaz-Arab-Latn-2012	alalc-kazakh.pdf (2012 version, "No earlier versions")	42
alalc-kur-Arab-Latn-2012	alalc-kurdish.pdf (2012; Arabic column of the Hawar/Cyrillic/Arabic trilingual table)	83
alalc-lad-Hebr-Latn-2025	alalc-ladino.pdf (2025; Hebrew script)	86
alalc-jba-Hebr-Latn-2011	alalc-judeo-arabic.pdf (2011-01-31; Hebrew script)	92
alalc-zsm-Arab-Latn-2011	alalc-malay.pdf (2011; Malay in Jawi-Arabic script)	55
alalc-jav-Arab-Latn-2012	alalc-jawi-pegon.pdf (2012 Jawi/Pegon, successor of the Malay table)	58
alalc-snd-Arab-Latn-2024	alalc-sindhi.pdf (2024)	61
alalc-urd-Arab-Latn-2013	alalc-urdu.pdf (2013)	52
alalc-uig-Arab-Latn-2015	alalc-uighur.pdf (2015)	52
alalc-pus-Arab-Latn-2013	alalc-pushto.pdf (2013)	57
alalc-ota-Arab-Latn-1997	alalc-ottoman.pdf (undated; Birnbaum-based, 1997 system edition)	43

Moghol: source not in the collection (alalc-moghol.pdf does not exist) — not fabricated, per the work order.

Verification

• Embedded tests: 681/681 pass through the interscript-ts runtime (parser + converter + transliterate), run via a local runner over the 11 new maps.
• TS parse: npx -y tsx scripts/parse-isc.mjs ../interscript-ts → Parsed 420/420 .isc files.
• Ruby parse: Interscript::Isc::Parser.parse + DocumentBuilder.build over all 420 files against a fresh origin/main clone of interscript-ruby (parsanol 1.3.135) → Parsed 420/420 .isc files.
• git status: only the 11 new .isc files; no existing file modified.

Encoding decisions worth review

• Kurdish h/e (note 4). The hāʼ joining forms carry the H/E distinction. The map renders the heh presentation forms semantically (ﻫ ﻬ → h; ﻪ ﻩ → e; doachashmee forms accordingly, with final ﮫ → e per the ême example) and resolves plain codepoints by context (word-initial he and he before a dual-joining letter = h, else e). Adjacent plain heh pairs (E+H as in lehī) are irreducible from codepoints alone — shaped input romanizes exactly; the tests carry the document's own text-layer encodings.
• Kazakh he e/h homograph (alphabet rows 8 and 29): defaults to e, exercised by the document's own example; the h reading is lexical and documented.
• Unwritten short vowels. Urdu note 5, Pushto note 4, Malay and Sindhi tables state vowels must be supplied from a dictionary. The maps encode the letter values, the written marks, and the mechanical rules; examples whose romanization requires lexical vowels are carried as comments with the mechanical rendering noted (e.g. Urdu هم → mechanical hm, document ham).
• Ottoman rule 2 (conversion). The principle of conversion to modern Turkish orthography is lexical; the map is the letter-by-letter table per rule 9, with the vowel-sign machinery deferred to the Arabic/Persian tables as the document itself directs. The source document carries no edition date; the 1997 attribute follows the ALA-LC 1997 system edition.
• Pushto ṭ row. The 2013 source's romanization cell for ṭ is not legible in the extraction; the value ṭ follows the earlier ALA-LC Pushto tables and the universal Arabic-script reading (flagged in the map's notes).
• Undotted/dotted homographs (Kazakh he, Uighur he/ه, Ottoman گ written stroke-less as ك, ڭ written dot-less as ن, Malay ف as shorthand for ڤ): both codepoints are encoded with the table-row values; which reading a given token carries is lexical and noted in each map.
• Shaped text layers. Kazakh, Ladino and Kurdish tests carry the documents' presentation-form encodings verbatim (in Kurdish and Ladino they are semantically load-bearing); Malay and Uighur presentation forms carry no letter identity, so those maps operate on base codepoints and the tests are the logical transcriptions.
• Case. Kazakh rule 3, Ottoman/Urdu/Pushto rule on capitalization and the proper-noun capitals in several sources are writer conventions; the maps leave case untouched and the affected examples are commented rather than forced.
• Moghol is recorded as source-not-in-collection (above) and nothing else.