# I'm inventing a COMPLETE brand-new alphabet - how many letters do I need?

**URL:** <https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860>\
**Category:** Factual Questions\
**Created:** [January 25, 2011, 1:02am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860 "2011-01-25T01:02:05Z")\
**Posts on this page:** 20\
**Page:** 2

<div class="post-metadata">

**Author:** ![EvilTOJ](https://avatars.discourse-cdn.com/v4/letter/e/ee59a6/32.png) [@EvilTOJ](https://boards.straightdope.com/u/EvilTOJ)\
**Post date:** [January 25, 2011, 9:51am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/21 "2011-01-25T09:51:14Z")

</div>

Someone told you to describe something in your own words, and you’re taking it to the next step, aren’t you?

---

<div class="post-metadata">

**Author:** ![Uncertain](https://avatars.discourse-cdn.com/v4/letter/u/6a8cbe/32.png) [@Uncertain](https://boards.straightdope.com/u/Uncertain)\
**Post date:** [January 25, 2011, 12:30pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/22 "2011-01-25T12:30:36Z")

</div>

> [@Wogglebug](#):
>
> Prolly should have mentioned that I’m a professional linguist

“Prolly”?

---

<div class="post-metadata">

**Author:** ![CalMeacham](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/calmeacham/32/35_2.png) [@CalMeacham](https://boards.straightdope.com/u/CalMeacham)\
**Post date:** [January 25, 2011, 12:55pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/23 "2011-01-25T12:55:53Z")

</div>

The Shaw Phonetic Alphabet, like Unifon, uses 40:

> **[Shavian alphabet](https://en.wikipedia.org/wiki/Shavian_alphabet)**
>
> The Shavian alphabet (/ˈʃeɪviən/; also known as the Shaw alphabet) is an alphabet conceived as a way to provide simple, phonemic orthography for the English language to replace the difficulties of conventional spelling using the Latin alphabet. It was posthumously funded by and named after Irish playwright Bernard Shaw. 
> Shaw set three main criteria for the new alphabet. It should be: 
> The Shavian alphabet consists of three types of letters: tall, deep and short. Short letters are vowels, liq...

I don’t think there’s a single unequivocal answer. Henry Higgins, in Shaw’s play _Pygmalion_, certainly seems to suggest that there are many more sounds than a mere 40 characters would describe – and Shaw, of course, invented the Shavian alphabet. But you don’t want your alphabet to get too long and character-intensive. People have to learn and remember the damned thing, and extra letters are going to add significantly to reproduction and printing costs (especially in a pre-inkjet world)

---

<div class="post-metadata">

**Author:** ![Colophon](https://avatars.discourse-cdn.com/v4/letter/c/f05b48/32.png) [@Colophon](https://boards.straightdope.com/u/Colophon)\
**Post date:** [January 25, 2011, 6:08pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/24 "2011-01-25T18:08:38Z")

</div>

> [@CalMeacham](#):
>
> People have to learn and remember the damned thing, and extra letters are going to add significantly to reproduction and printing costs (especially in a pre-inkjet world)

Really? Surely having more letters available means you don’t have to use so many characters - e.g. instead of “sh” you could use the single character ʃ, and so on. Think of the Cyrillic character Щ, which requires four characters to transcribe in English (shch), and seven in German (schtsch).

Sure, you’d have to spend a bit more at the outset on casting more type, but in the long run you’d save lots of paper and ink.

---

<div class="post-metadata">

**Author:** ![CalMeacham](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/calmeacham/32/35_2.png) [@CalMeacham](https://boards.straightdope.com/u/CalMeacham)\
**Post date:** [January 25, 2011, 7:04pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/25 "2011-01-25T19:04:22Z")

</div>

> [@](#):
>
> Sure, you’d have to spend a bit more at the outset on casting more type, but in the long run you’d save lots of paper and ink.

It’s not just that — you’d have to have _lots_ more characters set in metal, and lots more drawers to hold them (an awful lot more if you have different sizes and typefaces). And it will cause much more complication when you go to keyboards and the like.  
Just revisit the questions, posted here many times, about how the Chinese managed with their movable type and keyboards.

---

<div class="post-metadata">

**Author:** ![Chronos](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/chronos/32/134_2.png) [@Chronos](https://boards.straightdope.com/u/Chronos)\
**Post date:** [January 25, 2011, 7:15pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/26 "2011-01-25T19:15:10Z")

</div>

What about letters with various accent marks and other adornments? If you used, say, an unadorned a, an a with an umlaut, an a with a circumflex, etc. to represent the various English sounds represented by a, would that count as one letter or several?

---

<div class="post-metadata">

**Author:** ![Sunspace](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/sunspace/32/1250_2.png) [@Sunspace](https://boards.straightdope.com/u/Sunspace)\
**Post date:** [January 25, 2011, 7:32pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/27 "2011-01-25T19:32:35Z")

</div>

> [@Chronos](#):
>
> What about letters with various accent marks and other adornments? If you used, say, an unadorned a, an a with an umlaut, an a with a circumflex, etc. to represent the various English sounds represented by a, would that count as one letter or several?

Depends on the language. In French, é and e are just variants of e. In Esperanto, ĝ and g are two completely-different letters.

So we’d have to make a decision.

---

<div class="post-metadata">

**Author:** ![Acsenray](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/acsenray/32/4519_2.png) [@Acsenray](https://boards.straightdope.com/u/Acsenray)\
**Post date:** [January 25, 2011, 7:36pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/28 "2011-01-25T19:36:13Z")

</div>

> [@Wogglebug](#):
>
> Somewhat grinning at Arnold for asking this question. Troublemaker!
> 
> You’d need symbols for all the pronunciation keys mentioned by, say the Websters.

You’re kidding, right? Those have been obsolete for decades. The International Phonetic Alphabet has what you need.

---

<div class="post-metadata">

**Author:** ![Arnold\_Winkelried](https://avatars.discourse-cdn.com/v4/letter/a/3d9bf3/32.png) [@Arnold\_Winkelried](https://boards.straightdope.com/u/Arnold_Winkelried)\
**Post date:** [January 25, 2011, 10:35pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/29 "2011-01-25T22:35:54Z")

</div>

> [@Uncertain](#):
>
> You’ll have the problem that different English speakers pronounce words differently. How, for example, will you spell “schedule”?

That is always a problem. If a word can be pronounced several different ways, then it can be written several different ways. All those sounds will correspond to the same word.

> [@Martian Bigfoot](#):
>
> Dunno if it’s relevant, bur if you’re looking for tiny languages, the Rotokas language, spoken on Papua New Guinea, apparently gets by with just 12 letters and 11 phonemes (why the extra letter? Pure extravagance, obviously.)

I’m not looking for the smallest working language with the least letters, I want to devise an alphabet that can cover all the sounds used in any possible language, with one letter per sound.

> [@Chronos](#):
>
> What about letters with various accent marks and other adornments? If you used, say, an unadorned a, an a with an umlaut, an a with a circumflex, etc. to represent the various English sounds represented by a, would that count as one letter or several?

In my scheme, that would count as several, if they are pronounced differently.

I took a look at the IPA, but if I follow my last rule, I would have to know how many different sounds the IPA can represent. That would be the number of letters in my alphabet. It’s not clear to me, from looking at the IPA chart, how many different symbols (plus combinations of more than one symbol, used to represent a single “sound”) are used in the IPA.

---

<div class="post-metadata">

**Author:** ![Derleth](https://avatars.discourse-cdn.com/v4/letter/d/b9e5f3/32.png) [@Derleth](https://boards.straightdope.com/u/Derleth)\
**Post date:** [January 25, 2011, 10:46pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/30 "2011-01-25T22:46:36Z")

</div>

> [@Arnold\_Winkelried](#):
>
> I’m not looking for the smallest working language with the least letters, I want to devise an alphabet that can cover all the sounds used in any possible language, with one letter per sound.

Then you have a harder task than real linguists, who limit themselves to the phonology that’s _actually_ used in human languages. To do what you’re proposing, you’ll need to do a comprehensive study of the _entire_ human vocal apparatus, from the diaphragm to the nasal sinuses, and figure out _every possible sound_ it can make.

I suspect a strong plurality of the sounds you discover (or derive from the principles of resonance, vibration, and airflow) would be alien to every human language; assigning letters to them would be eccentric, to say the least, and would increase the number of distinct glyphs required\* to the point where you’ll have trouble getting anyone to draw a comprehensive font.

\*(This assumes you insist on each distinct letter being implemented as a distinct glyph. You could have an inventory of base glyphs and an inventory of diacritical marks and rely on software to combine the two into finished glyphs. This is actually a workable solution, though the result is rarely ‘letter-quality’, as it were.)

---

<div class="post-metadata">

**Author:** ![Arnold\_Winkelried](https://avatars.discourse-cdn.com/v4/letter/a/3d9bf3/32.png) [@Arnold\_Winkelried](https://boards.straightdope.com/u/Arnold_Winkelried)\
**Post date:** [January 25, 2011, 11:01pm UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/31 "2011-01-25T23:01:29Z")

</div>

Let’s change it to this then:  
I’m not looking for the smallest working language with the least letters, I want to devise an alphabet that can cover all the sounds used in _currently existing languages_, with one letter per sound.

---

<div class="post-metadata">

**Author:** ![orcenio](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/orcenio/32/1036_2.png) [@orcenio](https://boards.straightdope.com/u/orcenio)\
**Post date:** [January 26, 2011, 12:04am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/32 "2011-01-26T00:04:32Z")

</div>

> [@Arnold\_Winkelried](#):
>
> Inspired by this thread:  
> [Things that the English language desperately needs](http://boards.straightdope.com/sdmb/showthread.php?t=594052)
> 
> in particular this link to a [Mark Twain quote](http://design.caltech.edu/erik/Misc/Twain_english.html), mentioned inside the above thread
> 
> I want to (take your pick)  
> a) invent a new language,  
> or  
> b) simplify the spelling of an existing language, for example English
> 
> My ground rule is:  
> I want one letter per sound that you can make. A one-to-one correspondance, not like the english confusion where c is sometimes the same as s, and sometimes the same as k, and where k is redundant with q, and where you need two letters for some sounds, like th. In my alphabet, the f sound will always be an f. The z sound will always be a z. ph will always be written f. etc…
> 
> I know that I’ll have to write down foreign words sometimes, like Chinese place names, or Hawaiian proper names, so I need to have enough letters that I can also represent the sounds found in any other spoken language.
> 
> How many letters will I need?

The number of glyphs you use all depends on what type of writing system you want to churn out.

You gots your standard **Alphabet** where each letter (grapheme/character/glyph/etc) corresponds, roughly, to a phoneme which are joined to create words. Most of Europe uses this system due to the Greeks and their ca-rayz system of writing. ~20-30 characters are normal.  
An example: abcdefghijklmnopqrstuvwxyz

Then you gots your **Abjad**. It’s basically the same \*\*Alphabet \*\*model, but you knock out all the vowels to make things more interesting. t’s knd fckd p sstm bt sm ppl lk t tht wy. See? A little more difficult then your standard alphabet since the full words are merely an image of how it sounds; _vowel intensive_ languages (like Native Hawaiian) would be rendered as garbage. Arabic uses this one; ~20-30 characters are normal too.  
An example: bcdfghjklmnpqrstvwxz

Then things get… different. Your standard **Abugida** is basically the **Abjad** writing system (without any vowels), but you take each consonant (~20-30) and create a corresponding character for each possible vowel. So take the number of consonants (20) and multiply that with the number of vowels (6) and you get ~120 characters. Devanagari (Hindi) uses this with ~1296 viable symbols.  
An example:  
ba be bi bo bu by  
ca ce ci co cu cy  
da de di do du dy  
…

Your standard \*\*syllabary \*\* is basically the same as your standard **Abugida** model where each written symbol represents a syllable, but there is no consonant/vowel multiplication goin’ on. So it’s much smaller then an **Abugida** , but it still needs to have a large bank of syllables to use and combine together so it will be larger than an **Alphabet**. ~40-90 is a good average number and Japanese Hiragana is an example of this.

Finally you gots yer \*\*logographs \*\*where each written symbol represents not a sound (Phoneme), but an idea (wetness, brightness, etc). Good examples are some hieroglyphs and Chinese. ~1000-the sky’s the limit man…

---

<div class="post-metadata">

**Author:** ![Chronos](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/chronos/32/134_2.png) [@Chronos](https://boards.straightdope.com/u/Chronos)\
**Post date:** [January 26, 2011, 12:48am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/33 "2011-01-26T00:48:01Z")

</div>

And of course, there’s variation within any of those categories, too. For instance, among the logographs, there’s variation in how representational the symbols are. Egyptian hieroglyphs are mostly little pictures of things, and often pictures of the specific things they represent, but Chinese characters, while they might have started out that way, have become stylized to unrecognizability in most cases. Plus, among the phonetic writing systems, there’s the question of the shape of the symbols and how they’re organized: In the Latin alphabet, for instance, f and v look completely different and are nowhere near each other in alphabetical order, despite having very similar sounds.

---

<div class="post-metadata">

**Author:** ![Arnold\_Winkelried](https://avatars.discourse-cdn.com/v4/letter/a/3d9bf3/32.png) [@Arnold\_Winkelried](https://boards.straightdope.com/u/Arnold_Winkelried)\
**Post date:** [January 26, 2011, 12:56am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/34 "2011-01-26T00:56:36Z")

</div>

> [@orcenio](#):
>
> The number of glyphs you use all depends on what type of writing system you want to churn out.
> 
> You gots your standard **Alphabet** where each letter (grapheme/character/glyph/etc) corresponds, roughly, to a phoneme which are joined to create words.

This is what I want: an **Alphabet**.

> [@Chronos](#):
>
> among the phonetic writing systems, there’s the question of the shape of the symbols and how they’re organized: In the Latin alphabet, for instance, f and v look completely different and are nowhere near each other in alphabetical order, despite having very similar sounds.

I don’t care if similar sounds have similar-looking letters, and/or if the letters are next to each other in the alphabet.

What I am looking for is how many different letters I will need in my alphabet.

---

<div class="post-metadata">

**Author:** ![orcenio](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/orcenio/32/1036_2.png) [@orcenio](https://boards.straightdope.com/u/orcenio)\
**Post date:** [January 26, 2011, 1:02am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/35 "2011-01-26T01:02:29Z")

</div>

> [@Arnold\_Winkelried](#):
>
> This is what I want: an **Alphabet**.  
> I don’t care if similar sounds have similar-looking letters, and/or if the letters are next to each other in the alphabet.
> 
> What I am looking for is how many different letters I will need in my alphabet.

I highly suggest an nice **Abjad** with the _vowel markings_ option added. It serves the same as an **Alphabet** , functions well in it’s pointed/unpointed forms, is much more standardizable and is very compact in script and word length.

---

<div class="post-metadata">

**Author:** ![Acsenray](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/acsenray/32/4519_2.png) [@Acsenray](https://boards.straightdope.com/u/Acsenray)\
**Post date:** [January 26, 2011, 1:06am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/36 "2011-01-26T01:06:28Z")

</div>

**Thelurkinghorror** told you in the first reply to your OP: “As of 2008, there are 107 letters, 52 diacritics, and four prosodic marks in the IPA.”

---

<div class="post-metadata">

**Author:** ![Superfluous\_Parentheses](https://avatars.discourse-cdn.com/v4/letter/s/8edcca/32.png) [@Superfluous\_Parentheses](https://boards.straightdope.com/u/Superfluous_Parentheses)\
**Post date:** [January 26, 2011, 1:17am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/37 "2011-01-26T01:17:18Z")

</div>

For English, or one of the other roughly European languages, you need about 20 to 140 glyphs corresponding to the basic sounds. Once you cross language borders you’ll probably run into lots of trouble though. Especially if your goal is to go world-wide. For example, some east-Asian languages also take into account the pitch, which is pretty much ignored in European languages.

---

<div class="post-metadata">

**Author:** ![Arnold\_Winkelried](https://avatars.discourse-cdn.com/v4/letter/a/3d9bf3/32.png) [@Arnold\_Winkelried](https://boards.straightdope.com/u/Arnold_Winkelried)\
**Post date:** [January 26, 2011, 1:32am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/38 "2011-01-26T01:32:51Z")

</div>

> [@orcenio](#):
>
> I highly suggest an nice **Abjad** with the _vowel markings_ option added. It serves the same as an **Alphabet** , functions well in it’s pointed/unpointed forms, is much more standardizable and is very compact in script and word length.

I would be loathe to give up my vowels as separate letters. As a matter of fact, I refuse to do so on principle.

> [@acsenray](#):
>
> **Thelurkinghorror** told you in the first reply to your OP: “As of 2008, there are 107 letters, 52 diacritics, and four prosodic marks in the IPA.”

I appreciate Thelurkinghorror’s informational post, but as I mentioned previously,

> [@Arnold\_Winkelried](#):
>
> I took a look at the IPA, but if I follow my last rule, I would have to know how many different sounds the IPA can represent. That would be the number of letters in my alphabet. It’s not clear to me, from looking at the IPA chart, how many different symbols (plus combinations of more than one symbol, used to represent a single “sound”) are used in the IPA.

Let’s assume I don’t want to apply diacritics to the letters in my alphabet. e.g. I do **not** want what we have in French, e and è and é. In my alphabet, those would be three different letters. How many letters would I need? I assume the answer is **not** 107 (letters) × 52 (diacritics) × 4 (prosodic marks) = 22256 letters. (but maybe it is!)

P.S. Of course, I’m not really going to go to the trouble of making my own alphabet. But **IF** I were going to, how many letters would I need?  
Or, to rephrase the question, the IPA was constructed to represent a certain number of sounds that exist in current languages. What is the total number of sounds represented in the IPA?

---

<div class="post-metadata">

**Author:** ![Arnold\_Winkelried](https://avatars.discourse-cdn.com/v4/letter/a/3d9bf3/32.png) [@Arnold\_Winkelried](https://boards.straightdope.com/u/Arnold_Winkelried)\
**Post date:** [January 26, 2011, 1:33am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/39 "2011-01-26T01:33:59Z")

</div>

> [@Superfluous\_Parentheses](#):
>
> Especially if your goal is to go world-wide.

**THAT** is my goal.

---

<div class="post-metadata">

**Author:** ![orcenio](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/orcenio/32/1036_2.png) [@orcenio](https://boards.straightdope.com/u/orcenio)\
**Post date:** [January 26, 2011, 1:47am UTC](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860/40 "2011-01-26T01:47:20Z")

</div>

> [@Arnold\_Winkelried](#):
>
> I would be loathe to give up my vowels as separate letters. As a matter of fact, I refuse to do so on principle.

Then the IPA would be your best bet, as they do leave the vowels in as letters (a memory costly, but nicely distinct, feature); while keeping the total alphabet length as concise as possible (as concise a global alphabet can be).

> [@](#):
>
> I appreciate Thelurkinghorror’s informational post, but as I mentioned previously,
> 
> Let’s assume I don’t want to apply diacritics to the letters in my alphabet. e.g. I do **not** want what we have in French, e and è and é. In my alphabet, those would be three different letters. How many letters would I need? I assume the answer is **not** 107 (letters) × 52 (diacritics) × 4 (prosodic marks) = 22256 letters. (but maybe it is!)
> 
> P.S. Of course, I’m not really going to go to the trouble of making my own alphabet. But **IF** I were going to, how many letters would I need?  
> Or, to rephrase the question, the IPA was constructed to represent a certain number of sounds that exist in current languages. What is the total number of sounds represented in the IPA?

What you want is some sort of…super- **Abugida**!

Not Vowels x Consonants = Letters but  
Letters (Vowels x Consonants?) × Diacritics × Prosodic marks = Full Character list!

You’re mad. That’s just inhuman! You try singing that **Alphabet** song, bub; you won’t end untill November. Illiteracy will sky-rocket due to the unrememberable **Alphabet** and the SDMB will be abandoned for real life companionship. We’d be back into the dark ages…

[Previous page](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860.md?page=1)

[Next page](https://boards.straightdope.com/t/im-inventing-a-complete-brand-new-alphabet-how-many-letters-do-i-need/568860.md?page=3)
