Icelandic
Icelandic read the way it has always been written
Icelandic has changed so little that a modern reader can still work through a fourteenth-century saga manuscript, and its everyday print still carries two letters, þ and ð, that the rest of Europe dropped centuries ago. Evidano reads both ends of that continuity, from the abbreviation-heavy vellum hand of the sagas to an ordinary newspaper page, without collapsing þ and ð into a p or a d.
Sample pages
Real pages from public collections, shown beside their transcriptions. Pages whose transcription is still being checked are marked.
Árni Magnússon Institute for Icelandic Studies (Stofnun Árna Magnússonar í íslenskum fræðum), Reykjavík, GKS 1005 fol., f. 76r, c. 1387–1394 · Source · Public domain · Jón Þórðarson and Magnús Þórhallsson (scribes); image via handrit.is
Transcription being verified
Printed diplomatically in G. Vigfússon and C. R. Unger’s edition, Flateyjarbok: en Samling af norske Konge-Sagaer (Christiania, 1860), vol. I, in the opening chapters of Haralds saga hárfagra; matching this folio’s exact lines to that edition’s pagination is still being checked.
What the model is told to watch for: Produce a diplomatic transcription in Icelandic.
Transcription
The cover shown, transcribed on Icelandic Wikisource and validated to its highest proofreading status; the transcription supplies the author’s name and title as printed, including the words this particular copy’s shelfmark sticker now covers.
JÓHANN SIGURJÓNSSON GALDRA-LOFTUR LEIKRIT Í ÞREM ÞÁTTUM REYKJAVÍK ÚTGEFANDI: ÞORSTEINN GÍSLASON 1915
Transcription from Blaðsíða:Galdra-Loftur (1915).djvu/1, via Icelandic Wikisource (Public domain).
Nationalmuseet (National Museum of Denmark), 19 August 1899 · Source · Public domain
Transcription being verified
No proofread transcription of this issue has been published; the National and University Library of Iceland’s newspaper portal, timarit.is, carries an OCR layer for Ísafold that a transcription could be checked against.
What the model is told to watch for: Produce a diplomatic transcription in Icelandic.
Source · Public domain · Photograph by Jabbi
Transcription being verified
A short public plaque rather than an edited text; no separate published transcription exists, and the wording can be read directly from the photograph.
What the model is told to watch for: Produce a diplomatic transcription in Icelandic.
From vellum to newsprint, one language throughout
Iceland was settled from Norway and the British Isles in the ninth century, and the sagas describing that settlement were written down from the thirteenth century onward in a Gothic bookhand on vellum: rounded minuscule letters, heavy use of suspension strokes and superscript letters, and a Tironian ⁊ for og. Because Icelandic pronunciation and grammar have changed comparatively little since then, a modern Icelander can read Njáls saga or Egils saga in something close to the original spelling, an experience with no real equivalent in English or the other Nordic languages. The great vellum codices, among them Flateyjarbók and the Poetic Edda’s Codex Regius, were carried to Copenhagen for study in the seventeenth and eighteenth centuries, above all by the collector Árni Magnússon, and most were returned to Iceland during the twentieth century; they are now held together at the Árni Magnússon Institute for Icelandic Studies in Reykjavík.
Two of the manuscript hand’s letters never left the printed language. Þ (thorn) and ð (eth), both inherited from the same runic-Latin mixture that gave Old English its thorn and eth, still open and close ordinary Icelandic words today, the only place in the modern world where either survives in everyday print. Icelandic also keeps six vowels marked with an acute accent, á, é, í, ó, ú and ý, each a distinct vowel rather than a stressed one, plus ö; a model that treats the accent as decoration and strips it turns one word into another. Manuscript abbreviation thinned out only slowly: early printed books, from the 1540 New Testament to the 1584 Guðbrandsbiblía, still carried suspended endings and the ⁊, and it is really the eighteenth century before Icelandic print settles into the spelled-out alphabet used now.
Everyday Icelandic material for transcription is just as continuous. A nineteenth-century newspaper sets its news and advertisements in the same þ-and-ð alphabet a reader uses today; a hand-lettered information plaque on a farm building or a village pool carries a full paragraph of prose, capitals and all; and modern Icelandic cursive, taught to a distinct national model, still gives þ a descender and ð a crossed ascender that are easy for a generic model to misread as p, b or d.
Why ordinary OCR struggles here
Þ, ð and the accented vowels
Þ and ð sit outside the Latin alphabet most models are tuned on, so þ is read as p or y and ð as d or o; the six acute-accented vowels fare little better, since á, é, í, ó, ú and ý mark distinct vowels rather than stress, and dropping the accent turns one word into a different one entirely.
A bookhand built on suspension
The saga hand marks a missing letter or syllable with a stroke above the line, a raised letter, or a shape borrowed from Latin scribal shorthand, and the same stroke can resolve differently depending on the word around it. Reading it means recognising the abbreviation system, not just decoding individual letterforms.
Two columns, dense and continuous
Saga pages are ruled into two tight columns with no paragraph breaks, chapter headings picked out only in a red rubric, and a coloured initial the only signal that a new section has begun. Column order and rubric placement have to be read correctly before a single word is transcribed, or lines from opposite columns end up stitched together.
A language readable across eight centuries
Because Icelandic has barely changed, a model can mistake a genuine medieval spelling for a typing error, or make the opposite mistake and silently modernise a saga’s own forms to match today’s dictionary. Either error erases exactly the evidence a philologist is trying to read.
Who works with this material
Saga editors and philologists
Scholarly editions of the sagas and eddic poems are built folio by folio from the vellum, distinguishing what the scribe wrote from what a later editor supplies. A transcription that keeps suspension marks, rubrics and folio breaks in place gives an editor a usable first pass rather than a normalised paraphrase.
Genealogists and local-history projects
Iceland’s patronymic naming and small population make parish registers, censuses and local newspapers unusually rich genealogical sources, but only if þ, ð and the accented vowels in a name are kept exact; Þórðardóttir and Thordardottir will not match in a search.
Museums and heritage groups
Small museums, swimming pools and farms across Iceland carry hand-lettered or printed information boards explaining their own history, often the only record of a detail nobody wrote down elsewhere. Digitising them usefully means transcribing the sign as it stands, not correcting its spelling or dropping its diacritics.
Newspaper and archive digitisation
Iceland’s nineteenth- and twentieth-century press ran for decades in the same dense, small-type column layout, and a project indexing it needs þ and ð recognised as themselves rather than folded into t, y, p or d, or a search for þingmaður will miss half of its own archive.
Getting the best transcription
Say which letters must survive
List þ, ð and the six acute-accented vowels in the prompt and ask for them to be kept exactly as printed. That one step is what stops þing becoming ping and Þórunn becoming Thorunn.
Set an abbreviation policy for manuscript pages
For saga vellum, decide whether suspended letters should be expanded silently, expanded with the supplied letters marked, or left as the scribe’s stroke. State it once and the same rule holds across a whole codex.
Name the folio’s two columns
Ask for the left column to be transcribed in full before the right, and for rubrics and initials to be flagged rather than merged into the surrounding prose. Getting the reading order right the first time avoids sentences stitched together from opposite columns.
Test on the most damaged item first
Before running a full batch of signs, newsprint or vellum, start with whichever item is faded, cracked or worn at the edges. If illegible stretches are marked rather than guessed there, the cleaner pages in the same batch need only a spot check.
Frequently asked questions
- Will þ and ð come through correctly?
- Yes. Both are recognised as their own letters rather than variants of p, y or d, on manuscript vellum and on a modern printed sign alike, and the same applies to the acute-accented vowels á, é, í, ó, ú and ý.
- Can it read a saga manuscript’s abbreviations?
- It recognises suspension strokes, raised letters and the Tironian ⁊ as abbreviations rather than stray marks. You choose whether they are expanded silently, expanded with the supplied letters marked, or left as the scribe wrote them.
- Will it modernise an old spelling to match today’s Icelandic?
- Not unless you ask for it. Because Icelandic has changed so little, it is easy for a model to blur a genuine medieval or nineteenth-century spelling into the modern form; the default keeps the page’s own spelling so the date and hand remain visible.
- Can it keep a saga folio’s two columns in the right order?
- Yes, if the layout is described in the prompt. Reading the left column fully before the right, and marking rubrics and initials rather than folding them into the text, keeps the sense of the page intact.
- Does it work on photographed signs as well as scans?
- Yes. A photographed information plaque or shop sign is read the same way as a scanned page, allowing for the angle and lighting of the photograph, and a short sign with no separate published source is marked as such rather than left unmarked.
Related scripts and pages
Transcribe your Icelandic documents
Upload a saga folio, a newspaper page or a photographed sign, and check the first page for þ, ð and the accented vowels before running the rest.