Arabic · c. 1600–1950
From a copyist’s naskh to a journalist’s crossed-out draft
Between 1600 and 1950 the same language was written in two very different ways: a careful naskh for books, and a fast ruqʿah or half-ruqʿah for everything else, in which dots merge, vowels vanish and a whole word can shrink to a zigzag. The presses added a third layer, with numerals, running heads and two languages sharing a page.
Sample pages
Real pages from public collections, shown beside their transcriptions. Pages whose transcription is still being checked are marked.
Bibliothèque universitaire des langues et civilisations (BULAC), Paris, MS.ARA.67, 1770 · Source · Public domain
Transcription
The whole text block, from the entry al-mursala to al-insān al-kāmil, proofread on Arabic Wikisource against this same scan. Words the manuscript writes in red are the headwords, which stand at the start of each line below; the isolated nūn is the section letter, written in red and centred in the transcription’s source.
المرسلة أن يشهد رجلان في شيء ولم يذكرا سبب الملك إن كان جارية لا يحل وطؤها وإن كان دارا يغرم الشاهدان قيمتها الإمامية وهم الذين قالوا بالنص الجلي على إمامة علي رضي الله عنه وكفروا الصحابة رضي الله عنهم وهم الذين خرجوا على علي رضي الله عنه عند التحكم وكفروه وهم اثنا عشر الف رجل كانوا أهل صلاة وصيام وفيهم قال النبي عليه السلام يحقر أحدكم صلاته في جنب صلاتهم وصومه في جنب صومهم ولكن لا يتجاوز إسمانهم تراقيهم ن الانزعاج تحرك القلب إلى الله بتأثير الوعظ والسماع فيه الانصداع هو الفرق بعد الجمع بظهور الكثرة واعتبار صفاتها انتباه زجر الحق للعبد بإلقاءَات مزعجة منشطة إياه من عقال الغرة على طريق العناية به آنية تحقق الوجود العيني من حيث رتبته الذاتية الإنسان هو الحيوان الناطق الإنسان الكامل هو الجامع بجميع العوالم الإلهية والكونية الكلية والجزئية وهو كتاب جامع الكتب الإلهية والكونية فمن حيث روحه وعقله كتاب عقلي يسمى بأم الكتاب ومن حيث قلبه كتاب اللوح المحفوظ ومن حيث نفسه كتاب المحو والإثبات فهو الصحف المكرمة
Transcription from صفحة:كتاب التعريفات (IA MS.ARA.67).pdf/33, Arabic Wikisource (Public domain).
Bibliotheca Alexandrina, 1828 · Source · CC BY-SA 4.0 · Rania al-Bahara
Transcription being verified
Early runs of the gazette have been reprinted in Egypt, but no free transcription of this sheet is available; the Arabic column is legible enough to be read directly.
What the model is told to watch for: Expect formal Naskh in copied texts and compact Ruqʿah or mixed cursive in notes, letters, registers, and colophons. In rapid hands, dots may merge, move, or disappear, short vowels are normally absent, and connected groups may be reduced to angular skeletons; distinguish ه, ة, ي, ى, ن, and final loops from context and stroke direction. Note abbreviated invocations, honorific formulae, Ottoman-style numerals, catchwords, marginal ownership notes, seals, and changes between formal main text and informal annotations.
Bibliothèque universitaire des langues et civilisations (BULAC), Paris, RES MON 8 185, 1869 · Source · Public domain
Transcription being verified
Sīrat ʿAntar exists in several nineteenth-century printings whose wording differs; a transcription of this page would have to be checked against this printing rather than a modern edition.
What the model is told to watch for: Expect formal Naskh in copied texts and compact Ruqʿah or mixed cursive in notes, letters, registers, and colophons. In rapid hands, dots may merge, move, or disappear, short vowels are normally absent, and connected groups may be reduced to angular skeletons; distinguish ه, ة, ي, ى, ن, and final loops from context and stroke direction. Note abbreviated invocations, honorific formulae, Ottoman-style numerals, catchwords, marginal ownership notes, seals, and changes between formal main text and informal annotations.
Transcription being verified
Early volumes of al-Hilāl have not been proofread on Arabic Wikisource; the scan carries uncorrected machine-generated text only.
What the model is told to watch for: Expect formal Naskh in copied texts and compact Ruqʿah or mixed cursive in notes, letters, registers, and colophons. In rapid hands, dots may merge, move, or disappear, short vowels are normally absent, and connected groups may be reduced to angular skeletons; distinguish ه, ة, ي, ى, ن, and final loops from context and stroke direction. Note abbreviated invocations, honorific formulae, Ottoman-style numerals, catchwords, marginal ownership notes, seals, and changes between formal main text and informal annotations.
Transcription being verified
The draft is reproduced without a transcription; Zaydān’s published works are on Arabic Wikisource and could be used to identify the passage.
What the model is told to watch for: Expect formal Naskh in copied texts and compact Ruqʿah or mixed cursive in notes, letters, registers, and colophons. In rapid hands, dots may merge, move, or disappear, short vowels are normally absent, and connected groups may be reduced to angular skeletons; distinguish ه, ة, ي, ى, ن, and final loops from context and stroke direction. Note abbreviated invocations, honorific formulae, Ottoman-style numerals, catchwords, marginal ownership notes, seals, and changes between formal main text and informal annotations.
What Arabic looks like in these three centuries
Manuscript copying went on unchanged well into the nineteenth century. A scholar’s copy is written in a plain naskh, ruled to a fixed block, with headwords and key phrases picked out in red, a catchword at the foot, and a colophon giving the copyist and the date. Alongside it stands the everyday hand. Ruqʿah, codified in the Ottoman chancery and taught in schools across the Arab provinces, is small, tightly joined and written with a very short pen movement: the teeth of sīn flatten to a straight line, the loops of medial hāʾ and ʿayn close up, and the two dots of tāʾ or yāʾ become a single dash or disappear. Notes, letters, registers, marginal comments and drafts are written in it or in something less tidy still.
Print arrives in Arabic-speaking countries in force in the nineteenth century. The Būlāq press at Cairo, founded in the 1820s, printed the government gazette al-Waqāʾiʿ al-Miṣriyya as well as editions of the classics; presses in Beirut, Istanbul and Malta followed, and by the 1890s there were magazines like al-Hilāl and al-Muqtaṭaf with running heads, framed columns and page numbers. Early printed pages keep habits inherited from the manuscript: a rule around the text block, a catchword or quire signature at the foot, partial vocalisation of difficult words, and the same rubrication done with a heavier type. Government printing was often bilingual, with Ottoman Turkish and Arabic in parallel columns on the same sheet.
What makes this material awkward is the mixture. Eastern Arabic numerals run left to right inside a line that runs right to left, and a date, a page number or a quantity therefore reverses direction in the middle of a sentence. European personal and place names appear in Latin letters or in an Arabic transcription that varies from page to page. Drafts are covered in deletions, insertions written above the line and words crossed through so heavily that only the shape below the ink survives. Registers and forms put all of this into ruled boxes, where a heading may be printed and the entry written by hand.
Why ordinary OCR struggles here
Ruqʿah drops the information the reader needs
In a fast hand the dots move, merge or go missing, short vowels are absent, and final ه, ة, ي and ى all end in a similar hook. What separates them is the direction the pen was travelling and the grammar of the sentence, not the shape on the page, which is exactly what shape-based recognition cannot see.
Numerals that run the other way
A line of Arabic runs right to left; the digits inside it run left to right. Dates, page references, sums of money and years of the hijrī and Gregorian calendars therefore break the reading order several times in a paragraph, and a transcription that gets the order wrong turns ١٢٤٤ into a different number.
Two languages, one alphabet
An official gazette may set Ottoman Turkish beside Arabic in the same script, and a scholarly page may quote Persian. The letters are the same and the languages are not, so the columns have to be separated before either can be read; merged, they produce a text that is neither.
Drafts, deletions and insertions
An author’s working page is a record of second thoughts: words struck through, replacements squeezed above the line, whole passages ruled out. A useful transcription shows what was deleted and what replaced it rather than silently printing the final state or, worse, reading the cancelled words as part of the sentence.
Who works with this material
Newspaper and periodical digitisation projects
Runs of al-Hilāl, al-Muqtaṭaf, al-Ahrām and dozens of provincial papers exist as page images with no usable text. Getting the columns, the running heads and the figures right is what makes a run searchable by name, place and date.
Archivists working with Ottoman provincial records
Court registers, endowment deeds and petitions from the Arab provinces are written in a fast hand and often mix Arabic with Ottoman Turkish formulae. A transcription that keeps the two apart and preserves the dates is the basis for any index of parties, places and sums.
Editors of modern Arabic literary papers
Authors’ drafts, letters and notebooks from the nahḍa onwards survive in private and institutional collections. What an editor needs is not a clean reading text but a record of the deletions and insertions, since those are the evidence for how a piece was written.
Family and local historians
Identity papers, land certificates, school records and letters from the late Ottoman and mandate periods are the everyday sources for tracing a family. They are short, formulaic and written in a hand few people now read fluently, which is exactly where a first transcription helps most.
Getting the best transcription
Say whether the page is set or written
A printed page and a ruqʿah note need different instructions. For print, ask for the columns, running heads and footlines to be labelled; for handwriting, warn the model that pointing is unreliable and that words may be reduced to a skeleton.
Fix the treatment of numerals
Decide whether Arabic-Indic figures should be kept as ١٢٣ or converted to 123, and whether hijrī dates should be left as written. State it once: numbers are where reading order goes wrong, and a consistent rule makes the errors easy to spot.
Ask for deletions to be shown
On a draft, request that struck-through words are transcribed and marked as deleted, and that insertions above the line are placed where they belong with a note that they were added. That turns a messy page into something an editor can actually collate.
Separate the languages before reading
On a bilingual gazette or a register with Ottoman formulae, ask for each column or passage to be identified by language first and transcribed second. Test on a single sheet before running a whole volume, because a column that is misassigned once will be misassigned everywhere.
Further reading on this hand
The palaeography guides the in-app selector points to for this domain, if you want to check a transcription against the standard references.
- Arabic Paleography
This Hill Museum & Manuscript Library (HMML) course spans Arabic manuscript writing from the ninth through the twentieth century. Its later modules contrast carefully executed Naskh with increasingly casual hands found in colophons, notes, and modern-era manuscripts. The examples demonstrate positional letterforms, variable pointing, ligatures, vocalization, abbreviations, and changes in formality within a codex. That chronological range makes the course useful for Ottoman-era Arabic materials combining formal copying with rapid documentary annotation.
Frequently asked questions
- Can it read ruqʿah when the dots are missing?
- Often, because the missing dots are recoverable from the word and the phrase rather than the shape. Readings that depend on context alone can be flagged, so you can check the names and numbers, which are the places where context gives no help.
- How are Arabic-Indic numerals handled?
- They are transcribed in the order they appear on the page, not in the order the surrounding right-to-left text runs, and you can choose whether to keep the original figures or convert them. Dates can be left in the hijrī form the document uses.
- What about a page that is half Ottoman Turkish?
- Each column or passage can be labelled with its language and transcribed separately. The two share an alphabet but not a vocabulary, and keeping them apart also lets a Turkish specialist take over the Ottoman half without untangling it from the Arabic first.
- Is printed nineteenth-century Arabic easier than manuscript?
- Not always. Early type is uneven, ligatures are cast as single sorts, the ink bleeds through thin paper, and the frames and footlines confuse column detection. It helps to say what the page furniture is so that a quire signature is not read as the last line of the text.
- Can crossed-out words be recovered from a draft?
- Sometimes. Where a single stroke runs through the word the letters below are usually still readable and can be transcribed with a note; where the ink is scribbled solid the passage is marked as deleted and illegible rather than guessed at.
Related scripts and pages
Transcribe your Ottoman-era Arabic pages
Upload a manuscript leaf, a gazette or a draft, choose Ottoman-Era Arabic Naskh and Ruqʿah Documents, and set the rule for numerals first.