The spelling alphabet used by aviation and the military writes Alfa without a ph and Juliett with two ts. Both look like errors and neither is: the words were respelled so that French and Spanish speakers would pronounce them as intended, since ph is not a familiar spelling for an f sound and a final single t would be silent.
That is the whole design principle. The alphabet was built for a radio channel with interference and listeners who did not share a first language, and every choice in it serves being understood rather than being written.
Why those particular words?
Because they survived testing. The set was assembled and revised through the 1950s by testing candidate words with speakers of several languages under poor listening conditions, and words that were confused with each other were replaced.
The words are deliberately long and dissimilar. A syllable count of two or three gives a listener more to recognise than a single syllable, and no two words in the set share a stressed vowel and a consonant frame.
That is why substituting your own words is a worse idea than it sounds. An improvised alphabet reuses common sounds and loses exactly the property that makes the standard one work.
Where does it get used?
Anywhere a string has to survive being said aloud.
| Situation | Why |
|---|---|
| Reading a reference number | avoids M and N, B and D |
| Dictating an email address | avoids the vowel confusions |
| Radio and phone with noise | designed for exactly this |
| Confirming a booking code | one letter wrong is a lost booking |
The digits have their own conventions in aviation, where nine is said as "niner" so it cannot be heard as the German nein, and three is often said without the th sound that some speakers do not have.
For ordinary use the letters do most of the work. The pairs that cause almost all misunderstandings are M and N, B and D and P, and S and F, and the alphabet separates each of them completely.
How does braille encode letters?
In a cell of six dots, arranged in two columns of three. Six positions each either raised or flat gives 64 combinations including the empty cell, which is enough for the alphabet, punctuation and a set of indicators.
It was devised by Louis Braille, who began work at fifteen and published the system in the 1820s, adapting a military code intended for reading messages at night without light. The military version used twelve dots, which was too wide to read under one fingertip — reducing it to six is what made it work.
There are two levels in ordinary use. One spells everything letter by letter; the other uses contractions, where a single cell stands for a common word or letter group, which makes the text considerably shorter and takes longer to learn.
Why does that matter for a converter?
Because a letter-by-letter conversion is not what a fluent reader expects. Producing uncontracted braille is straightforward and correct; producing contracted braille requires knowing the contraction rules of the specific language.
It also matters physically. Braille is meant to be embossed at standard dot spacing, so a visual representation on a screen is a transcription rather than a usable document — useful for checking and for learning, not for reading by touch.
Capital letters and numbers are marked by preceding indicator cells rather than by different shapes, which is why a braille transcription is longer than the text it came from even before contractions.
What about play languages?
They are transformations rather than codes, and they follow rules simple enough to apply in your head. Moving the leading consonant cluster to the end and adding a fixed ending is the whole of the best-known one.
They are reversible only when the rule is unambiguous, which it usually is not. A word starting with a vowel has no consonant cluster to move, so the convention has to invent one — and different speakers invent different ones.
That makes them a good demonstration of why a real cipher specifies its edge cases and a playground rule does not.
What makes a code easy to say?
Avoiding the characters that collide. A reference containing both O and 0, or both 1 and I and l, will be misread by someone eventually — which is why identifier alphabets designed for humans leave those characters out entirely.
Length matters less than composition. A longer code from a safe alphabet is easier to convey accurately than a short one containing an ambiguous pair, and it is easier to check by reading back.
Questions people ask
Does it work over a bad phone line? That is what it was built for. It is most useful exactly when it feels most unnecessary.
Is there one official spelling alphabet? The aviation one is the international standard. Police and military services sometimes use their own.
Should I use it on a customer call? For anything alphanumeric, yes. It sounds formal and it prevents the callback.
Is braille the same in every language? No. The letter assignments and contractions are language-specific.
Can braille be read from a screen? Only through a refreshable display with physical pins. A visual rendering is for sighted transcription.
Say it so it survives the channel. The NATO phonetic alphabet converts a string into words that cannot be misheard, the braille translator transcribes text into the six-dot cells, and the pig latin translator handles the playground rule.