Spelling in your conlang

[Linguist Wannabe]

How regular is the spelling in your conlang? Is it possible to predict the pronunciation of a word just from seeing it written? How about vice versa? Do different groups of speakers use different spelling systems (e.g. British vs. American?

I've heard that in Middle English, spelling was much less standardised. Is that the case today?

[Drydic]

English spelling isn't standardized. It's fossilized.

[Thry]

Irlandic is like French, you can read what is written (save for some stress if you don't write the accents), but you mostly can't spell from just the sounds. It's fossilized too. Different dialects have different pronunciations and different informal spellings, but the formal register has it standarized. Some examples:

/phonemic/ [standard] [North]

fablad ("spoken") /faw.ˈlaw/ [fɔ.ˈlaw] [fɔ.ˈlɔ]
ácid ("acid") /ˈɛ.jew/ [ˈɛ.je] [ˈɛ.jo]
vida ("life") /vew/ [vew] [fe]
fod ("he fucks") /fow/ [fow] [fu]
deia ("of the") /ˈdjə/ [djə] [dʒə] (sometimes spelt dya; gya or ge in the North only, always colloquially)
die ("day") /ˈdi.ə/ [ˈdi.ə] [di]
decembre ("december") /di.ˈɛm.bɾə/ [ˈdjɛm.bɾə] [dʒɛm]
battya ("may it beat") /ˈbɛ.tjə/ [ˈbɛ.tʲə] [bɛtʃ(ə)]
pállid ("pale") /ˈpa.lew/ [ˈpa.le] [ˈpa.lo]
terra ("soil") /ˈtɛ.rə/ [ˈtɛ.rə] [tæɛ̯]
perdre ("to lose") /ˈpɛɾ.dɾə/ [ˈpɛɾ.dɾe] [pæɛ̯ð]
ser ("to be") /seɾ/ [seɾ] [sɛe̯]
morg ("I die") /moɾg/ [moɾg] [mɔo̯ɣ]
mors ("you die") /mɔɾs/ [mɔɾs] [mɑɔ̯s]
morya ("may he/she die") /ˈmo.ɾjə/ [ˈmo.ɾʲə] [ˈmo.ɾʲə]~[ˈmu.ɾʲə]
aspeils ("mirrors") /ɐs.ˈpeʎs/ [ɐs.ˈpeʲʎts] [spilts]
castanyas ("chestnut") /kɐs.ˈtɛɲs/ [kɐs.ˈtɛʲɲs] [kəs.ˈtɛntʃs]
brylyar ("to shine") /bɾi.ˈʎaɾ/ /ˈʎe/ [bɾi.ˈʎaɾ] [bɾi.ˈʎe] (same in both dialects, first form isn't used in the North colloquially)

[Dewrad]

Dravean, by and large, is fairly phonemic in its orthography (it was only reformed about sixty years ago): there aren't any instances where the pronunciation of a word cannot be predicted with full assurance from the orthography. However, predicting the orthography from the pronunciation of a word is slightly trickier, given that a few sounds may be written using different graphemes, particularly the vowels. For example, pretonic [ u ] might be written u or o:

utail [uˈtaiɫ]
compagna [kumˈpaɲǝ]

A final [ŋ] could be written n, nt or nd:

cantand [kǝŋˈtaŋ]
quènt [ˈkɛŋ]
can [ˈkaŋ]

Similarly, it is difficult to predict whether a spoken will correspond to a written ei or i:

òmnei [ˈɔmni]
taci [ˈtaʃi]

[finlay]

My plan with Sentalian was to create a bunch of dialects in which different words are pronounced wildly differently, but tend to keep the original phonemic spelling of the mother dialect. The problem is I ended up making something that's not aesthetically pleasing. It has sonority-hierarchy breaking coda clusters just for the sake of getting high tone vowels in a dialect when voiced plosives are deleted.

The last time I really did much with it was about two years ago, and I had by that time six daughter dialects. For an example, the word dvogl, meaning ten, was pronounced [dʋogl] in the mother dialect, and respectively:
[dʋól], [tfəgl], [vól], [dwəkl], [dvɔɣ], [dɹoːl]
in the daughters. So you've got varying degrees of what is deducible from the spelling alone – but like French or something, it'd be difficult to go backwards and get the spelling from the pronunciation, especially with the third dialect, which has the heaviest cluster reduction.

With my other conlang, Yaufulti, which I have kept up a little bit more recently, the spelling is "phonemic" but I have several synchronic phonological processes in the language, making pronunciation sometimes hard to predict (however, when I tried speaking it, I didn't like the result, so I currently want to revise it completely). In the older version, anyway, there was consonant harmony of manner of articulation, then intervocalic voicing, then palatalisation, to the point where tati, tazi /taʒi/, tasi and tali were all pronounced [taːdʒi] and tani is the very similar [tãːdʒi], and then there's a triplet of sati, saki, sagi /saɣi/ came out as [saːʝi]. When I redo it I'll keep some of the mergers but not all, I think.

[Izambri]

fablad ("spoken") /faw.ˈlaw/ [fɔ.ˈlaw] [fɔ.ˈlɔ]
ácid ("acid") /ˈɛ.jew/ [ˈɛ.je] [ˈɛ.jo]
vida ("life") /vew/ [vew] [fe]
decembre ("december") /di.ˈɛm.bɾə/ [ˈdjɛm.bɾə] [dʒɛm]
castanyas ("chestnut") /kɐs.ˈtɛɲs/ [kɐs.ˈtɛʲɲs] [kəs.ˈtɛntʃs]

[Thry]

haure ("have") /ˈaw.ɾə/ [awɾ] [ɑɔ̯]
dubda ("doubt") /dow/ [dow] [du]
cobd ("elbow") /kow/ [kow] [ku]
secur ("safe") /sə.ˈkuɾ] [sə.ˈkuɾ] [skuː]
cerc ("near") /θɛɾk/ [sɛɾk] [fæɛ̯k]

there's no going back

c'mon, my L1 is Spanish so I'm repressed with our ridiculous phonemicity.

[Click]

Jkeurng has a 2500 years outdated spelling that reflects the native script.

jkeurng /tʂʰèŋ/ [ʨʰɛ̀ŋ]
gorj /kuè/ [kʷɞ̀]
mayt /mái/ [mɜ́]
tyatn-gong /tʰjánkōŋ/ [ʨɑ́nɡōŋ]
kbork /kʰuō¹/ [kʷʰō]

The spelling is hideous, of course, so I prefer to use a phonemic spelling based on Mandarin, where these example words would be spelled chèng, guè, mái, tiángong and guō.

[Thry]

Jkeurng has a 2500 years outdated spelling that reflects the native script.

holy crap

[Nortaneous]

Arve's writing system has been fossilized for about a thousand years, but in that time, it lost all its clusters and turned pronouns in preverbal position to affixes and then to initial mutations. It's usually possible to predict sound from spelling, although stress is unpredictable and some words are irregular, but it's not at all possible to predict spelling from sound. And there are three major dialect groups, plus the upper-class, conservative standard.

en sattes 'I am' [ˈr̥ʌʔɪs] [ʂʌjiɕ] [r̥ʌ̰i̯s] [r̥ɨ̰ːs]
jald 'be correct' [jɛɔ̯ts] [jæu̯ts] [jøə̯ts] [vjɞːs]
Arve 'Arve' [ˈhɛu̯wɪj] [ˈhæu̯wɪj] [ˈhɛu̯wɪj] [ˈɞːwi]
vöntus 'alcohol' [ˈⱱyø̯r̥ʊs] [ˈbɥiʂʊs] [ˈⱱyə̯r̥ʊs] [vyːʂ]
steck 'grain' [çʌk] [ɕɛk] [çʌʔ] [çɛx]
ächs 'snake' [hɛç] [hæɕ] [hɛç] [aç]
ettei 'noise' [hɪˈtʰʌi̯] [hɪˈtɛi̯] [hɪˈtʰʌi̯] [çtsɨː]
nävönd 'rock' [ˈnɛvʏr] [ˈnævʏr] [ˈnœy̯r] [nøː]
nögle 'branch' [ˈnʊi̯ɣɪj] [nwi] [ˈnʏi̯jɪj] [nøː]
scheng 'dog' [çʌʀ] [ɕɛŋ] [çʌʀ] [çɛʀ]

[Plusquamperfekt]

Miwonša has a very phonemic orthography:

http://conlang.wikia.com/wiki/Miwonsa#ORTHOGRAPHY

[Click]

Jkeurng has a 2500 years outdated spelling that reflects the native script.

holy crap

Oh, thanks.

I have one more conlang with an insane orthography, Skyran. This one is only 1400 years outdated. Enjoy.

skýræ /ˈʂɨrɛ/ [ˈʂɯ̄ʕɛ]
króðk /ˈqʰùkʰ/ [ˈqʰʊ̀kʰ]
skørkja /ˈʂɜtɕʰa/ [ˈʂʌ̀ʨʰɑ]
tjé /ˈtɕʰi/ [ˈʨʰī]
zænkýtr /sɛnˈkʰɨ́r/ [sɛ̃ˈkʰɯ́ʕ]

[Curlyjimsam]

Viksen spelling hasn't been updated for a few centuries and so some sound changes aren't reflected, e.g. <æ>, <r>, <l> and <h> aren't pronounced in final position. Some sounds can be written in more than one way reflecting different historical origins. There are also various more sporadic sources of irregularity.

Things get more complex when (recently loaned) non-native words are considered. These tend to be transliterated directly, but retain their original pronunciation. Thus for example <ss> in Atlian loanwords is pronounced /ʂ/, which is spelled <ś> or <sj> in native words - in those few native words in which <ss> occurs, it is pronounced /s.s/.

I would love to get around to working on an orthography for some conlang or other that is as non-standardised as Middle English's was, but haven't managed it yet.

[malavssi]

@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@

[Thry]

Arve's writing system has been fossilized for about a thousand years, but in that time, it lost all its clusters and turned pronouns in preverbal position to affixes and then to initial mutations. It's usually possible to predict sound from spelling, although stress is unpredictable and some words are irregular, but it's not at all possible to predict spelling from sound. And there are three major dialect groups, plus the upper-class, conservative standard.

en sattes 'I am' [ˈr̥ʌʔɪs] [ʂʌjiɕ] [r̥ʌ̰i̯s] [r̥ɨ̰ːs]
jald 'be correct' [jɛɔ̯ts] [jæu̯ts] [jøə̯ts] [vjɞːs]
Arve 'Arve' [ˈhɛu̯wɪj] [ˈhæu̯wɪj] [ˈhɛu̯wɪj] [ˈɞːwi]
vöntus 'alcohol' [ˈⱱyø̯r̥ʊs] [ˈbɥiʂʊs] [ˈⱱyə̯r̥ʊs] [vyːʂ]
steck 'grain' [çʌk] [ɕɛk] [çʌʔ] [çɛx]
ächs 'snake' [hɛç] [hæɕ] [hɛç] [aç]
ettei 'noise' [hɪˈtʰʌi̯] [hɪˈtɛi̯] [hɪˈtʰʌi̯] [çtsɨː]
nävönd 'rock' [ˈnɛvʏr] [ˈnævʏr] [ˈnœy̯r] [nøː]
nögle 'branch' [ˈnʊi̯ɣɪj] [nwi] [ˈnʏi̯jɪj] [nøː]
scheng 'dog' [çʌʀ] [ɕɛŋ] [çʌʀ] [çɛʀ]

A bit frivolous to call any of those conservative xD

How did you make the intermediate steps for nd > r?; ng > R?

[Nortaneous]

A bit frivolous to call any of those conservative xD

Well, Midland is the most conservative of the three... not that that's saying much, admittedly. There'll probably be some conditional vowel mergers present in Standard but not in Midland or vice versa, but I haven't gotten there yet. (The real reason is that I made Standard first, made the diphthongs saner, and then decided to split the two vowel systems into different dialects.

How did you make the intermediate steps for nd > r?; ng > R?

Nasal + stop clusters can pick up trilled release. Happened in Nias: mb nd > mbʙ ndr > ʙ dr, written as <mb ndr>. Same thing happened in Arve, except also to velar clusters; I figured it'd back to uvular, since there's no such thing as a velar trill. (mb nt nd nk ng > v r̥ r χ ʀ; mp isn't attested yet and probably won't be)

[Izambri]

haure ("have") /ˈaw.ɾə/ [awɾ] [ɑɔ̯]
dubda ("doubt") /dow/ [dow] [du]
cobd ("elbow") /kow/ [kow] [ku]
secur ("safe") /sə.ˈkuɾ] [sə.ˈkuɾ] [skuː]
cerc ("near") /θɛɾk/ [sɛɾk] [fæɛ̯k]

These ones seem more natural to me. Pronouncing intervocalic or final <bd> as [w], I mean.

Why cerc [fæɛ̯k] in Northern? Does <c> before <e> = [f]?

[Thry]

Those <-bd> were always final, though (-a being an illusion), what you can see as [w] is actually the earlier /d/. A lot of unexisting vowels are kept in the end, cf. 1st person singulars, which look like in Catalan, despite being more like in Valencian: io amo /jo am/. Though intervocally you can get it too; /dow.ˈde/ [du.ˈde] dubder "to doubt", but from the earlier /b/.

Not merely <e>, it's just that initial /θ/ is realized as [f] in the North Dialect - so that's ce ci. Non-initially it's [j], [s] or ellided depending on context (intervocally, after consonants, word-finally).

In the Standard Dialect it's /θ/ realized as [s] or [j], the distribution being as described but with [s] filling in for [f] except in very formal or solemn contexts where [θ] is kept. The [j] actually can already be considered /j/ (sound change) for all intents and purposes in any dialect, [s] can too. So:

Cels ("Heaven") /θɛls/ [sɛls] [fɛls]
accéder ("to access") /ɐk.ˈθe.də/ [ɐk.ˈse.də] [ək.ˈsed]
renovatión* ("renewal") /rə.no.və.ˈjon/ [rə.no.və.ˈjon] [rə.no.vi.ˈjon] *that -t- is just earlier /θ/, in medieval texts you can find -açión and -açyón
inice ("beginning") /i.ˈni.ə/ [i.ˈni.ə] [i.ˈni]

[GrinningManiac]

Well in-universe none of my languages would be written with any real-world alphabet but because writing word documents with a made-up font would be insanely tedious I tend to emphasise the difference between different branches of language families by having fun with spelling as well as phonology.

My favourite at the moment is Oumish - which is basically Manx in terms of how much sense it outwardly makes.

[Bristel]

Spelling in Proto-Takayo can be pretty phonemic, but there are quite a lot of exceptions for compound words and affixed words.

taraka ("light, brightness") /taraka/ [taraka]
kagu ("to give") /kaᵑgu/ [kaᵑgu]
nigi ("mother") /niᵑgi/ [niᵑgi]
paga ("father") /paᵑga/ [paᵑga]
kizauko ("---") /kiⁿzauko/ [kiⁿzako]
porituba ("---") /poriuba/ [porituᵐba] <-- in this case, epenthetic /t/ is inserted and is shown in the orthography.

[ol bofosh]

Alahithian: extremely phonetically faithful spelling.

ethey /ˈe.ʈej/
om /om/
bethaw /ˈbe.ʈaw/
daraw /ˈda.ɾaw/

[jal]

Kotanian is almost completely phonemic (with very littel allophony, so it's also rather phonetic), its romanization following the native script. For Alubetah I chose a phonetic romanization, as the language's extensive allophony would make a phonemic ortography rather useless (and, given the specific allophonic rules, in need of stress indications). Sajiwan (an carribean English creole) is fairly phonemic, although historical sound changes have obscured the obviousness somewhat (e.g. /wi/ > /y:/). In general, the languages I devise are linguistic exercises, without too much background. I therefore find it easier to device phonemic/phonetic ortographies.


JAL

[Toko]

The Zelsian Mik script is fairly fossilised. If you know the rules, you can probably predict the sound of a word with fair accuracy, because while taken symbol-for-symbol the script would barely correspond to the sound of the modern Zelsian pronunciation, a mildly complex set of rules does account for most things. But, there are still cases where you simply have to memorise.

The romanisation I use for it has no connection to the conworld itself (i.e. they have no knowledge of any places that would use a Latin script), so instead of trying to follow the native script it focuses almost solely on pronunciation aside from a few exceptions, like <sz>, which is just /z/, is written to reflect a similar combination in the native script.

[Qwynegold]

My alphabet arranges letters in block form. Associating letters to sounds is regular, but arranging letters in blocks is not. Most words over two syllables long have several ways of being written, depending on how the blocks are arranged. The sandard arrangement of each word is sometimes irregular and arbitrary. School students memorize the standard way of writing words. "Mispelled" words can still be read, but with more difficulty.

Alphabet Examples:

What writing materials/tools were used for the traditional script?

[malavssi]

What writing materials/tools were used for the traditional script?

@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@