Hey, Want to Learn Some Ottoman Turkish Writing?
fun and games with various Perso-Arabic alphabets, including some app recommendations
When I see something written in a language I don’t understand, I feel like I’m missing out. As an example, just look at this beautiful picture of a horse:
I love the smoothness of the lines and curves, both in the drawing and the calligraphy. And those colours! A beautiful Bay with the black points, yet also the yellow/red/white patterning1 like the way they draw extinct horse ancestors.
Now, for someone who grew up on a farm, I’m not an overly horsey person; I have only one scar on my body resulting from a horse-related injury, and that happened many years ago. I can tell, though, that whoever drew that horse really knew what they were looking at. You can even tell the breed - Arabian - from that arched neck, the raised tail, and the rather fierce look in the horse’s eye; the horse that scarred me was a Thoroughbred, which apparently inherited its hotblooded temperament from the Arabian in its breeding stock.
The picture comes from the Kitāb al-bayṭarah ( كتاب البيطرة “Veterinary Book”).2 It was apparently drawn almost exactly 800 years ago in August 1223, by a guy called Aḥmad ibn ‘Atīq al-Azdī (أحمد بن عتيق الأزدي). The best copy online seems to be in the Qatar Digital Library, but disappointingly there is only one more illustration in addition to the above.3 Although to my view the horse looks to be in great shape, the picture above is actually illustrating the horse’s defects. I’m so curious as to what defects are noted - don’t you wish you could read that Arabic script just a little?
The horse picture is entirely annotated in Arabic, specifically the Naskh calligraphic style. This is simultaneously one of the oldest styles of Islamic calligraphy, and also the basis for many modern electronic Arabic fonts.
Of course Ottoman Turkish, which was heavily influenced by Persian and Arabic, was also written using a Perso-Arabic script. Briefly, the Persian alphabet was derived from Arabic, but with four additional consonants. Ottoman Turkish then adds one more (the “muted kef” ڭـ with the three dots above, pronounced like the ng in English sing) for a total of 33 consonants. For simplicity, I’m just going to refer to them all as “Arabic” alphabets from now on.
Although the modern Latin-based Turkish alphabet is much more suitable for spelling Turkish words, the swap from Ottoman script had the effect of cutting Turks off from much of their history and literature. Here’s a recent article lamenting this loss, which also notes the beginnings of a solution: automated transcription of handwritten Ottoman documents using machine-learning tools such as Akis.
(Not many details on Akis available yet at that link, but the paper is said to be “forthcoming”. Seems they are training their own model, so as to be free to release it publicly, rather than relying on commercial LLMs)
The current Turkish government seems keen to support a revival of Ottoman Turkish. Quoting the Akis article:
This rise in interest has also been influenced by politics, with current President Recep Tayyip Erdogan a longtime advocate of neo-Ottomanism. In 2014, for example, a debate on whether to teach obligatory Ottoman language lessons at high school surfaced. While the education minister at the time was in favor of offering Ottoman as an optional subject, President Erdogan was supportive of compulsory classes, saying, “Whether they like it or not, Ottoman will be taught and learned in this country.” However, almost 10 years later, the teaching of Ottoman in high schools has still not come to fruition, perhaps because there aren’t enough teachers with sufficient knowledge of the lost language.
This comes up frequently in the historical dramas I like to watch, where the inclusion of archaic words and writing adds a bit of old-timey flavour. Here’s an example (Youtube link) from Kuruluş: Osman (ATV). Osman’s son Alaeddin (played by Yaman Çınar Balcı) has been kidnapped, and he writes a message in his own blood, warning his father that the arranged prisoner exchange is a trap:
Helpfully for the viewers, who aren’t expected to be able to read it, the bloody message is read aloud in the dialog: “gelmeyin” (“do not come”). Proudly, I can now confirm that it does look to be written exactly as pronounced in Modern Turkish; given that dialog prompt I can make out the five consonants n-y-m-l-g كلمين (remember Arabic scripts are written right-to-left) with no vowels written. There is a bit of a smear on the left of the circular “mim” consonant that could be something additional, and a weird hooked flourish at the end of the word (curving up from bottom left) that I don’t understand. It could possibly be Alaeddin’s way of signing his name - check out the similar flourish in his brother Orhan’s official signature in the picture below.4
So how did I get to this extremely basic level of Arabic reading ability? As usual, with apps from the app store. Read to the end of the article for recommendations, and you too can annoy your family and friends by pausing TV shows while you puzzle through bloody clues written in Arabic. First though, lets look at a more difficult example of Ottoman Turkish from TV.
I’ve always wondered about the accuracy of the archaic writing used on the TV shows; since almost noone reads Ottoman Turkish anymore, the producers needn’t worry that the audience will spot any mistakes. Obviously, the word “gelmeyin” written in blood above uses meaningful letters, but who knows if the precise wording corresponds to the way such things would have been expressed in the year 1300. What about the longer examples of ancient languages, such as in the screenshot below? Is this real Turkish?
The very first thing that strikes me when I look at the screenshot is the two “muted kef” characters ( ڭـ ) with the three dots, visible in the middle of the letter. We are off to a great start; it definitely looks to be Ottoman Turkish! Balance that out with the probably anachronistic appearance of an exclamation point in the final line; I don’t actually know when or where those were invented, but the internet tends to agree that the year 1300 would be about a hundred years too soon.
Do the TV producers have linguists working for them, who can write a letter in Ottoman Turkish or Greek whenever the script demands it? Or is it just gibberish text pasted in, like the Ottoman version of “Lorem Ipsum…”? A third possibility could be that a linguist wrote them a sample one time, and it’s been endlessly copied and re-used since. Let’s see if we can distinguish between those possibilities.
As context for the letter above, the season finale of Kuruluş: Osman was meant to be conveying a very significant (fictionalised) point in history. Here Sultan Mesut, last of the Great Selçuks, accepts that he can no longer count the lands occupied by Ottoman tribes as part of his empire. Conquered by the Mongols, his own light is fading, and Osman’s is brightening; he announces (Youtube link) that he is sending various Islamic scholars to Osman where they will be of more use, and that from now on, the evening “hutbe” sermons are to be read in Osman’s name. The latter is a very significant transfer of power at that time in history; in fact it was one of the two requirements for declaring yourself Emperor.5
So does the letter in the screenshot say anything about scholars or sermons? Time for a hint - lets ask Google. The translation we get is pretty much non-existent, since it thinks the language is Persian, but a few Persian-derived words come through ok:
“Scholars of knowledge in Islam” - now we are getting somewhere. Lets swap to the Persian transcription and see which words Google has recognised there.
I always say that machine transcription is way ahead of machine translation, but the results here surprised even me. Google’s transcription is remarkably good, considering it’s looking through my cheap phone camera, pointed at a screenshot of a language it doesn’t really understand, hand-written sideways on a piece of leather. Here is Google’s effort, mistakes and all:
رو عرى لتتونجا قريبه ده
ار شده اولمجق فتجلى في د
شكله به فکر را اعدى كلام ده
سجاق در سکر سکا عروس
، ، قاضی و عالمار کونده حکما
علم في اسلام بي جمله جهانه
يلياسك ! عزك مبارك اولسند
Let’s zoom in on the last four lines, with the sentence that talks about the scholars of Islam. Now the word used on the TV show to mean “scholars” is “ulema” علما and I’m almost certain that that word is present in the line third from the bottom. It’s in the phrase “kadı ve ulema” meaning “(Islamic) judges and scholars”:
Google hasn’t picked it up because the scribe made and corrected an error on the original; they’ve written “âlim” عالم at first (which is the singular form) and then changed it to “ulema” علما (the plural form) by overwriting the last two letters. Google reads the corrected word as a hybrid of the two: عالمار which it can’t make sense of.
The phrase that Google is reading as “scholars of knowledge in Islam” is in the second line from the bottom. The word “Islam” ( اسلام ) is a really easy one to spot, since it has that unusual “ـلا” character that represents “l” followed by “a”. The word before “Islam” is smudged at the end, but it definitely begins with the letters علم, which is the root meaning “knowledge” that the word for “scholars” is derived from:
I think though, that “knowledge of Islam” is probably the better translation here.
To cut a long story short, here is a better transcription of the last couple of sentences of Sultan Mesut’s letter. I’ve fixed some things where it seemed like Google had made an obvious error, but the stuff I wasn’t sure about I left as is:
And here is my transliteration of each word into faux Latinised Ottoman Turkish, followed by a translation into English. Without vowels its hard to know exactly what the scribe intended, and I have no idea about correct tenses and declension and what have you, so I won’t try to make proper sentences. As with the “scholar/knowledge” example above, I just figured out the main root word that’s implied by each cluster of consonants, and got a sense of the intended meaning from that. Corrections gladly welcome!
سڭا Sana “To you”
مدرس mudarras “teacher/educated”
قاضی و علما kadı ve ulema “Islamic judges and scholars”
کوندره gönder() “I send”
جکم ا cakma? <??>
علم في اسلام ilim() Islam “knowledge of Islam”
جمله cümle “all”
جهانه cihana “to the world”
ياياسڭ yayasın “spread; distribute”
غزك مبارى اولسن gazak mübari olsun “may your holy war be blessed”
Note: I’m not sure if “mudarras” is an authentic Ottoman word; it’s just an Arabic word I found that almost fits. It’s from the same root as “madrasa” (“school”), but without vowels etc. I can’t tell if it’s a noun or an adjective here. If it were a noun, then the Sultan could be sending “teachers, judges and scholars”; if an adjective then the judges and scholars are just particularly well educated.
Wow, that was a difficult yet satisfying task. I won’t try to translate the rest of the letter, since I think we have enough to conclude that indeed, Kuruluş: Osman employs an Ottoman Turkish expert who can write them a bespoke document matching the scriptwriters needs. Pretty impressive!
Some final thoughts before I give those Arabic alphabet app recommendations:
Knowing a little Turkish, but almost zero Arabic, I conclude that Arabic looks much harder than Turkish for English speakers to learn. This is even ignoring the different alphabet and the difficult consonant sounds - just the grammar by itself looks much more complicated than Turkish. If you don’t believe me, click this link to the Arabic word “mudarras”, and then click the two grey boxes labeled “declension of noun/adjective” to expand them. Since some of this complexity was definitely present in Ottoman Turkish (see above in the way that the plural of “âlim” is “ulema”) but is no longer, it underscores just how effective the Turkish language reforms have been over the past 100 years.
Arabic handwriting transcription is eminently possible with the technology we (i.e. Google) have already. The Akis project has every chance of success, and I wish them the best of luck. Kolay gelsin!
Arabic Alphabet and Writing App Recommendations
There are some very easy things about Arabic, and some very difficult things. Fortunately, the alphabet is one of the easy ones. It’s beautifully phonetic, and unlike English pronunciation, Modern Standard Arabic tends to follow it’s own rules (although there are rather a lot of them). This consistency is mostly because the Muslim holy book, the Qur’an, is written in Classical Arabic, and divine punishment is said to fall on those who would change the meaning of its words. Modern Standard Arabic was developed as a revised and simplified version of Classical Arabic, so that everyone could agree on a literary standard for formal use (newspapers or official documents etc.), regardless of which Arabic dialect you might speak day-to-day.
Most of the major language apps will teach spoken MSA, with little attention paid to reading. The Drops app, for example, only displayed the Romanised spelling by default (e.g. “hello” is “marḥaban”); I had to find the setting for the Arabic script and turn it on, which I feel downplays something very important - not just for reading, but for understanding the language itself. Fortunately the basics of the alphabet are pretty similar across Arabic dialects, Ottoman Turkish, and Persian, and there are quite a few standalone apps that concentrate on learning Arabic characters. Helpfully, the fact that all Muslims are encouraged to learn Classical Arabic in order to read the Qur’an in its definitive edition means that many Arabic learning resources are provided free, or very low cost. This brings me to my first recommendation.
1. Arabic Alphabet for Beginners
This app is exactly what I wanted: clear pictures of each letter, with numerous examples of pronunciation in combination with various vowel sounds, and in whole words. Since the Qur’an is the only place where you will tend to see Arabic vowels written down, the major language apps will omit them. This app shows the vowels marks in every case, which is extremely helpful for understanding how pronunciation matches the spelling.
The free version of Arabic Alphabet for Beginners covers all 28 standard Arabic consonants, with a separate lesson for each followed by some well-designed multiple choice test questions. The second half of the app will cost you a tiny sum for access, and has some more advanced concepts,6 plus a heap more examples. You could get by just with the free version as an introduction to the alphabet, but the second half is very good value for money and is highly recommended.
Just as a benchmark, here’s how you would expect to perform trying to read some random Arabic text. The picture is of a jar of tahini (sesame paste) from my cupboard.
If you had only completed the free version of Arabic Alphabet for Beginners, you would be fine to transcribe those first four consonants طحينـ n-y-ḥ-ṭ (although you wouldn’t fully understand why there had to be a “y” in the middle), but that final one ـة (looking like a “ha” with two extra dots on the top) would be a complete mystery.
The paid version of the app would teach you the purpose of the “y” (it elongates the “i” sound), but you would still be a little confused about the ـة; under the app’s Classical pronunciation you might expect the final syllable to be pronounced “-tin”.
What the app doesn’t tell you, the internet will. “Tahini” comes from the colloquial pronunciation where the “-at-” sound is mostly disappearing. And under Classical rules it’s pronounced “-a” at the end of an utterance. Like I said, the alphabet itself is easy compared to grammar and pronunciation.
(“Tahini” is a particularly tricky example though: those first two consonants طحـ are really quite difficult for English speakers to pronounce, and then the grammar issue at the end is just icing on the cake)
2. Write It! Arabic
The thing with ubiquitous touch-screens and AI is that handwriting recognition should be more advanced than it is. As far as I know, Arabic handwriting input is not yet supported by Microsoft Windows at all. Google and Samsung both seem to be doing better with it, though.
As usual for this substack, lets pause to laugh good-naturedly at Google’s handwriting translation. If you look at internet comments a lot, you’ll quickly learn how to express laughter in Arabic: you type the letter ه repeatedly (it’s pronounced “ha”). This results in an aesthetically pleasing string looking something like petals along a twig: هههههههههه . Google converts these into English as strings of “hahaha”, of variable length. The curious part for me, is that its not a strict transliteration; the number of ه does not match the number of “ha”. In effect, Google is translating rather than transliterating: attempting to match the hilarity expressed by the two strings, rather than their length. Fascinatingly, the Google Translate AI has settled on a conversion rate for calculating humour content between the two languages. Smoothly incrementing the number of “ ه “ in Arabic returns a variable amount of English “ha”, until you get to 13 in a row where a sudden threshold is crossed; “ ههههههههههههه “ is invariably translated as “LOL”.
Anyway, back to the Write It! app. As the name suggests, this app will train you to write Arabic consonants on your phone screen with your finger, in word-initial, -middle or -final positions. This is followed by a timed reading/writing test on what you’ve just learned. The base app is free, but purchase gets you some customisable review settings, and removes some extremely intrusive advertising (if you’re playing the free version, its best to play with the sound off).
3. Arabic Alphabet Writing
The next-best competitor to Write It! Arabic is called Arabic Alphabet Writing. The main reason why I prefer the former is that Write It! has grid lines for you to write on. Without these I found it hard to appreciate the difference between د and ر and ل, for example. Also, Arabic Alphabet Writing is limited to the “single character” form of each consonant, not the joined-up cursive versions. Arabic Alphabet Writing still has some pretty good character recognition and a more fluid and aesthetically pleasing calligraphic style, and I would definitely try it again if I were interested in some of their other available languages.
This app is definitely set at the easiest difficulty level of the three, and it’s the one you should install first if you’re at all uncertain about your level of commitment to learning an Arabic alphabet. Whichever you choose though, just remember you’re in it for the long haul; do a little bit each day and don’t think about whether you’re improving or not. If you’re practicing regularly, you’re making progress!
If you want a horse like this for your collection in Red Dead Redemption 2, the closest might be the Warped Brindle Arabian. As well as looking great, it’s one of the fastest horses in the game.
Kitāb al-bayṭarah كتاب البيطرة Azdī, Aḥmad ibn ʿAtīq أزدي، أحمد بن عتيق [tail] (6/238), British Library: Oriental Manuscripts, Or 1523, in Qatar Digital Library <https://www.qdl.qa/archive/81055/vdc_100023486700.0x000007>
In that Mirador viewer, scroll to the left and the pictures are on pages [22av] and [62av]. Easily located if you’re scrolling quickly through the thumbnails at the bottom, just stop at the thumbnails that are double the normal width.
These “Tuğra” seals are interesting enough for a full post on their own. In lieu of that, this site shows how the style of Orhan’s original tuğra was continued and modified over time, becoming the signature style for the subsequent emperors.
The other being to mint your own coins, as I discussed here:
The advanced concepts covered are “madd” (indicating long vowels), “shaddah” (doubled consonants) and “tanwīn” (unwritten <n> sounds on the ends of words).