Alphanumerenglish

Kragen Javier Sitaker, 2015-04-06 (updated 2016-07-27) (6 minutes)

Arabic has some phonemes that don’t have a reasonable corresponding phoneme in Latin languages. A common approach when writing Arabic with Latin letters is to use European digits for the missing phonemes, as in “3arabawy”. This has the great advantage over X-SAMPA that it doesn’t deprive you of the uppercase/lowercase distinction entirely — although the digits themselves, in ASCII, don’t show the distinction visibly, you can still distinguish between “3arabawy” and “3ARABAWY”, or between “a7a”, “A7a”, and “A7A”. The European digits are assigned to sounds represented by Arabic letters that physically resemble them.

This seems like a great idea for English, too! English has a few more phonemes than the Latin alphabet, and consequently has to force some letters to represent multiple different phonemes, but its traditional orthography worsens the problem by wasting some of the Latin letters (c, q, and x) on redundant assignments. A rough phonological inventory of English follows, with a corresponding set of alphanumeric codes.

| Latin | IPA           | Alphanumerenglish |
| a     | æ a ɑ eɪ ə    | 4 a 2 8 e         |
| b     | b             | b                 |
| c     | k ʃ tʃ        | k c tc            |
| d     | d ɾ           | d d               |
| e     | ɛ i           | 3 i               |
| f     | f             | f                 |
| g     | g dʒ ʒ        | g dj j            |
| h     | h             | h                 |
| i     | aɪ ɪ i        | ai 1 i            |
| j     | dʒ ʒ          | dj j              |
| k     | k             | k                 |
| l     | l             | l                 |
| m     | m             | m                 |
| n     | n ŋ           | n 6               |
| o     | u ʊ ɑ ɔ oʊ æʊ | u 5 2 0 o au      |
| p     | p f           | p f               |
| q     | k             | k                 |
| r     | ɹ ɚ           | r er              |
| s     | s ʃ z         | s c z             |
| t     | t θ ð d ɾ     | t x q d d         |
| u     | u ʌ ju        | u 7 yu            |
| v     | v             | v                 |
| w     | w hw          | w hw              |
| x     | ks            | ks                |
| y     | j i ɪ aɪ      | y i 1 ai          |
| z     | z ʒ           | z j               |

In a few cases, I’ve taken the liberty of giving a diphthong a spelling that isn’t quite what you’d deduce logically, in the interest of making it easier to write: /oʊ/ is “o” rather than “o5” or “ou”, because I think /o/ doesn’t occur in outside of that diphthong in English, and likewise /eɪ/ is assigned a single digit “8”; /æʊ/ is “au” rather than “4u” or “45”, because although [au] is not the central instance of the /æʊ/ class, it’s certainly a valid pronunciation of it; and /aɪ/ is “ai” rather than “a1” for the same reason.

The mnemonics for the digits and other nonstandard pronunciations are as follows:

| 0 | ɔ  | looks like an O                             |
| 1 | ɪ  | looks like an I                             |
| 2 | ɑ  | no clue, sorry                              |
| 3 | ɛ  | looks like an ɛ, backwards                  |
| 4 | æ  | looks like an A                             |
| 5 | ʊ  | no clue, sorry                              |
| 6 | ŋ  | looks like the G in NG                      |
| 7 | ʌ  | looks like ʌ tilted a bit                   |
| 8 | eɪ | 8 is pronounced /eɪt/ in English            |
| c | tʃ | “c” is pronounced like standard “ch”        |
| e | ə  | looks like ə; also, the most common vowel   |
| q | ð  | no clue, and this one is super weird, sorry |
| x | θ  | no fucking clue, sorry                      |

In the interest of ease of typing and reading, I’ve tried to assign the less common sounds to digits, and especially digits that don’t look like anything. You might be able to get some benefit from swizzling around the assignment of 2 and 5, and maybe x and q too.

You might also get a readability improvement out of reassigning /ə/ to "a" rather than “e”, which would leave “e” free for /ɛ/, but would require reassigning /a/ to something else, such as “9” or “3”. /a/ alone is relatively rare in English, but “ai” and “au” are relatively common.

I’ve written a short Python hack to use eSpeak to convert traditional English orthography to Alphanumerenglish, not perfectly but with a tolerably low error rate. It turns out Alphanumerenglish is about 10% shorter.

Q7s wi ken rait i6gl1c perfektli fen3t1keli 1n 4lfenum3r1k w1q4ut nidi6 qe c1ft ki or qe los ev k8s d1sti6kcenz, werd sp8si6, 0r p76kcu8cen. 1t’s almost ridebel w164ut sp3cel tr8ni6, despait qe n3ses3ri k2mpremaizez 1n qi esainment ev gr4fimz te fonimz.

’Tw7z br1l1g, end qe slaiqi tovz
D1d gair end g1mbel 1n qe w8b.
Al m1mzi wer qe b0r0govz,
End qe mom r4xs autgr8b.

Biw3r qe dj4berw0k, mai s7n!
Qe dj0z qet bait, qe kl0z qet k0tc!
Biw3r qe dj7bdj7b berd, end c7n
Qe frumies b4ndersn4tc!

Hi t5k h1z v0rpel s0rd 1n h4nd.
L06 taim qe m86ksom fo hi s0t.
6en r3sted hi bai qe t7mt7m tri
End st5d e hwail 1n x0t.

End 4z 1n 7f1c x0t hi st5d
Qe dj4berw0k, w1x aiz ev fl8m
K8m hw1fli6 xru qe teldji w5d
End berbeld 4z 1t k8m!

W7n, tu! W7n, tu! End xru end xru
Qe v0rpel bl8d w3nt sn1ker-sn4k.
Hi l3ft 1t d3d, end w1x 1ts h3d
Hi w3nt gel7mfi6 b4k.

“End h4st qau slein qe dj4berw0k?
K7m tu mai armz, mai bim1c boi!
O fr4bdjes dei! Kelu, kel8!”
Hi tc0rdeld 1n h1z djoi.

’Tw7z br1l1g, end qe slaiqi tovz
D1d gair end g1mbel 1n qe w8b.
Al m1mzi wer qe b0r0govz,
End qe mom r4xs autgr8b.

Kemp3r q1s te qi X-SAMPA verjen:

twVz brIlIg, @nd D@ slAIDI toUvz
dId gAIr @nd gImb@l In q@ weIb.
al mImzi w@` D@ bOrOgovz,
@nd D@ moUm r4Ts aUtgreIb.

Te mai ai, et list, qi eks-s4mpe verjen 1z box l06ger end l3s ridebel ez n0rmel i6gl1c, ez w3l ez luzi6 qe degriz ev fridem ev k4p1telez8cen end sem p7nkcu8cen tu. Ev k0rs, X-SAMPA also h4z qi eb1ledi te r3prez3nt l86guedjez 7qer qen i6gl1c, almost ez izeli, hw1tc 4lfenum3ri6gl1c s4kr1faisez.

1ded bi jusfel te rait e w3b skr1pt te kenvert st4nderd i6gl1c te 4lfenum3ri6gl1c, m8bi jusi6 espeak -qx.

Topics