How to learn a new alphabet by reading

Learners of Russian, Arabic, Hindi or Japanese often meet a wall before they meet a single word: a page of letters that won’t turn into sounds. It looks like the hardest part of the language. It usually isn’t.

A script is a small, closed system. Greek has 24 letters. Russian has 33. Hebrew has 22, five of which change shape at the end of a word. Japanese hiragana has 46 basic characters. Compare that with the thousands of words you’ll learn in any language, and the script is the shortest list you’ll ever study.

What takes time is not learning the letters. It’s reading them fast enough that you stop seeing them. That part comes from reading, and it comes faster when you can hear what you’re reading.

A note on terms: strictly speaking, only some of these are alphabets. Arabic and Hebrew are abjads, which write consonants and leave most vowels to the reader. Devanagari is an abugida, in which each consonant carries a built-in vowel. Japanese kana are syllabaries. The approach below works for all of them.

Why reading teaches a script better than drills

Letter drills teach you to recognize one character at a time, slowly, out of context. Reading asks for something different: recognizing whole words, at speed, in sentences that mean something. The only way to practice that is to do it.

There is a deeper reason too. The reading researcher David Share described a “self-teaching” mechanism in children learning to read their first language (Share 1995, “Phonological recoding and self-teaching: sine qua non of reading acquisition”). Each time a reader sounds out an unfamiliar word successfully, they get a chance to store how that word is spelled, so it’s recognized faster the next time. Share was writing about children, and adults learning a second script are in a different position. But the logic is useful: every word you decode correctly is practice that pays twice, once for the letters and once for the word.

The catch is the word correctly. On your own, you can’t tell whether you read a word right. Audio can. Reading along with a narrator turns every sentence into a small, immediate check. (There’s more on why this works in reading while listening.)

The first two weeks

You don’t need a course to get through a script, but a little structure helps. Here is a plan built around 15 to 20 minutes a day.

Days 1–3: sounds first

Learn which sound each letter makes. Skip the letter names for now: you need to know that Russian Д sounds like d, not what it’s called. Work in groups of five or six letters, and write each one a few times by hand. Writing slows you down just enough to notice the details that separate similar letters.

Sort the letters as you go:

• Familiar: letters that look and sound like ones you know.
• False friends: letters that look familiar but sound different.
• New: everything else.

The false friends deserve the most attention, because your eyes will keep reading them the old way.

Days 4–7: real words, with audio

Start reading real text as soon as you know most of the letters. Choose the easiest level you can find; a new script is challenge enough without hard vocabulary on top. Play the audio and follow along. Then go back and read the same sentences aloud yourself, slowly.

Expect to be slow. A sentence a minute is not unusual at this stage. You’re not really reading yet, you’re decoding, and decoding is how reading starts.

Days 8–14: read first, then check

Now change the order. Read a short passage on your own first, sounding out each word. Then play the audio and check yourself. Your mistakes will start to have a pattern: one letter you keep reading as another, a vowel you keep missing. Those are the letters to review.

After two weeks of this, many learners can match the letters of an alphabet like Cyrillic or Greek to their sounds. Scripts with more characters or more shape changes, like Arabic or Devanagari, can take longer. Reading at an easy, unforced pace takes longer still, often months, and it comes from volume: more pages, more sentences, the same words met again and again.

Drop the romanization early

Transliteration (writing moloko for молоко) is useful for a day or two. After that it becomes a crutch, because every time you read the romanized form you skip the practice you need. If a textbook prints both, cover the romanization with your hand.

Script by script

Cyrillic: Russian, Bulgarian, Serbian and more

Cyrillic is used for Russian, Ukrainian, Bulgarian, Serbian, Macedonian, Belarusian, Kazakh, Mongolian and other languages, with some letters varying between them. For an English speaker, Russian’s 33 letters fall into three groups:

• Familiar: А, К, М, О and Т sound much as you’d expect.
• False friends: В is v, Н is n, Р is r, С is s, У is u, and Х is a throaty kh.
• New: Б, Г, Д, Ж, З, И, Л, П, Ф, Ц, Ч, Ш, Щ, Ы, Э, Ю, Я and a few more.

Try one now. Read as English letters, РЕСТОРАН is nonsense. Read with the false friends in mind, it says restoran.

Once the letters are in place, the real challenge in Russian is stress. Ordinary Russian text doesn’t mark which syllable is stressed, and unstressed vowels change their sound: an unstressed о sounds closer to a. Молоко (milk) is written with three o’s and sounds roughly like malakó. No chart will teach you the stress of every word. Audio will, one word at a time. If Russian is your language, start with learning Russian by reading.

Greek

Greek has 24 letters, and many will be familiar from mathematics and science: α, β, γ, δ, π, σ, Ω. Two habits make modern Greek easier than it first looks. It marks the stressed syllable with an accent on almost every word of more than one syllable, as in καλημέρα (kaliméra, good morning). And its spelling is consistent once you know the letter pairs:

• ου sounds like u, and αι like e.
• η, ι, υ, ει and οι all sound like i.
• μπ makes b and ντ makes d: μπανάνα (banána), ντομάτα (domáta).

Having several spellings for one sound makes writing Greek harder than reading it. For a reader, it’s a short list to learn. More in learning Greek by reading.

Hebrew

Hebrew is read from right to left. It has 22 letters, and five of them (כ מ נ פ צ) take a different form at the end of a word (ך ם ן ף ץ). A few letters look alike and are worth practicing in pairs: ב and כ, ד and ר, ה and ח and ת, ו and ז.

The bigger adjustment is vowels. Everyday Hebrew is written mostly without them. A system of dots and dashes called niqqud marks the vowels in children’s books, poetry and texts for learners. With niqqud, “peace” is שָׁלוֹם (shalom); without it, שלום. Start with niqqud and let it fade as your vocabulary grows. By then, the shapes of words you know will carry you. See learning Hebrew by reading.

Arabic

Arabic is also read from right to left. Its 28 letters join up within a word, and most of them change shape depending on whether they come at the start, in the middle or at the end. Six letters (ا د ذ ر ز و) never join to the letter after them, which is why you see small gaps inside words.

Dots matter. At the start or in the middle of a word, ب ت ث ن ي share one base shape and are told apart only by their dots: one below for b, two above for t, three above for th, one above for n, two below for y.

Like Hebrew, Arabic usually leaves short vowels unwritten, so the same three letters can be read in several ways. كتب can be:

• كَتَبَ kataba, “he wrote”
• كُتُب kutub, “books”
• كُتِبَ kutiba, “it was written”

Fluent readers choose from context. Beginners shouldn’t have to. Vowel marks help even experts: in a study of skilled adult readers of Arabic, reading Arabic and Hebrew, both vowels and context made reading significantly more accurate (Abu-Rabia 2001, “The role of vowels in reading Semitic scripts: Data from Arabic and Hebrew”). So start with fully vowelled text, so that every word can be read aloud exactly, and let the vowel marks fall away as you move up. Persian, Pashto, Urdu and Central Kurdish also use the Arabic script, each with letters of its own, and the same approach applies. See learning Arabic by reading.

Devanagari: Hindi, Marathi, Nepali

Devanagari works differently from the alphabets above. Each consonant carries a built-in a sound: क is ka. To write a different vowel, you add a mark to the consonant:

Written	Sound
क	ka
कि	ki
की	kī
कु	ku
के	ke
को	ko

Look at कि. The i mark is written before the consonant but pronounced after it. That’s the kind of detail that’s easy to read in a chart and easy to forget in a sentence, which is why reading real text matters so much.

Two more things to expect. Consonants with no vowel between them merge into combined shapes called conjuncts, like the स्त in नमस्ते (namaste). And in Hindi the built-in a is often silent at the end of a word and sometimes in the middle: कमल (lotus) is kamal, not kamala. Rules for this exist, but audio teaches it faster. See learning Hindi by reading.

The scripts of Bengali, Gujarati, Punjabi, Tamil, Telugu, Kannada, Malayalam and Sinhala belong to the same family, and so, further east, do Thai, Lao, Khmer and Burmese. The shapes differ, but the idea of a consonant plus a vowel mark carries over.

Japanese kana

Japanese uses three scripts together. Hiragana (46 basic characters) writes grammar endings and many native words. Katakana (also 46) writes words borrowed from other languages and adds emphasis. Kanji, characters that came from Chinese, write the core of most nouns and verbs.

Learn hiragana first, then katakana, then kanji gradually as you read. For most learners, kana take days or weeks; kanji are a project of years. That’s fine, because kana let you read real sentences long before you know many kanji.

Watch for three particles that break the usual rules: when they’re used as grammar words, は is read wa, へ is read e and を is read o. わたしは がくせいです (watashi wa gakusei desu) means “I’m a student.” See learning Japanese by reading.

And the rest

Korean Hangul was designed in the fifteenth century to be easy to learn, and it is: 24 basic letters, combined into syllable blocks. Georgian, Armenian, the Ge’ez script of Amharic and others each have their own logic. The method is the same everywhere: learn the sounds, then read real text with audio, a little every day.

Mistakes that slow people down

1. Studying letters for weeks before reading anything. Start reading as soon as you know most of the letters.
2. Choosing hard text. A new script plus new grammar plus unknown words is three problems at once. Read the easiest level available until the script feels automatic. (See how to choose your reading level.)
3. Reading in silence too early. Without audio, you’ll practice your mistakes until they feel right.
4. Leaning on romanization. It’s a bridge, not a road.
5. Skipping the vowels. In Arabic and Hebrew, begin with fully vowelled text.
6. Expecting speed too soon. Accuracy comes first. Speed follows from the number of pages you read.

How Lexo helps

Lexo is a reading app for iPhone and iPad, and several of its features suit this stage well:

• Hear any word. Tap a word to hear it on its own, with natural pronunciation, and to see what it means in that sentence.
• Read along. Every book is narrated, and the text highlights word by word as it’s read, so your eyes and ears stay on the same word.
• One sentence at a time. A focused mode shows a single sentence with its audio and translation, which is the right pace for the first days with a new script.
• Vowels where beginners need them. Arabic is Modern Standard Arabic, fully vowelled at A2 and B1. Hebrew has niqqud at A2 and B1. Japanese at A2 and B1 prefers kana to kanji above JLPT N3.
• Right to left, properly. Arabic, Hebrew and the other right-to-left languages read right to left, and their pages turn right to left.

Lexo covers about 80 languages, including Russian, Greek, Hebrew, Arabic, Persian, Hindi, Japanese, Korean, Thai and Georgian. A few, such as Ukrainian and Urdu, have text and glosses now, with narration coming later. Lexo is in beta and launching on the App Store soon. Get notified at launch.