What is with the dislike for digraphs when romanizing?

[Chagen]

Title says it all.

A lot of conlangers dislike digraphs, but I much, MUCH prefer them to diacritics when it comes to consonants (with vowels, on the other hand, I love absurd amount of diacritics--Vietnamese is one of my favorite natlangs for that reason). Diacritics on consonants look ugly as hell. Digraphs are so much prettier and nicer-looking--but just like with modern-day conworlds, I appear to be in the minority about this.

I just recieved a copy of the LCK (great book, by the way!) where Rosenfelder compares two orthographies for Verdurian, and said that <dh> was an uglier symbol than a <d> with circumflex (seriously, I cannot find this symbol anywhere). Personally, I think <dh> is a very pretty and nice digraph, while the d-with-circumflex was one of the ugliest letters I have ever seen.

So yeah. I just felt like asking this.

[Aurora Rossa]

I think diacritics have always appealed to me more because they seem much more exotic than digraphs. English has plenty of digraphs but no diacritics (apart from optionally in borrowings like "café" anyway). Diacritics do have the advantage of approaching a "one character, one phoneme" ideal better than digraphs, I would say.

[vampireshark]

When setting the orthography for Telèmor, I aimed to go for a certain aesthetic (Romanian/French hybrid). This is why there are digraphs for certain vowels (though many of the vowels are just reductions of varying clusters) and some diphthongs, but none for single consonant sounds.

[finlay]

I'd be wary of trying that, because exact phoneme-letter correspondence almost never happens (you have to match letters to sounds AND vice-versa - probably most languages have systems in which you can always perfectly read a word but can't predict the spelling if you only hear the sound (eg Spanish)).

But exoticism is pretty much hitting it on the head. People like things that aren't like English.

[Chagen]

Digraphs != English

Right now in my conalng Pazmat Wrlsjtvrdhya is an acceptable word yet I'm pretty sure that doesn't look that English-y.

[maıráí]

I use whichever works with the "feel" of a language, which is of course very subjective and probably not a good thing to go by.

Also, ease of typing, and whether or not there are enough diacritics in the first place to cover what I need covered.

[Thry]

I generally prefer digraphs over diacritics; I like how English lacks a written accent in fact (whereas my L1, Spanish, has the mark ´, and Portuguese has ^ ` ´ ~). But those are just tools, you can end with something beautiful or fugly either way.

[Izambri]

I like both, but my problem with diacritics is that they make certain consonants and capital vowels look uglier or ridiculous. Things like Ĝ, ħ, ĥ, Ị, Ł, Ŧ, ŧ, Ŵ or Ỳ just don't work.

Generally, I also dislike the use of more than one diacritic per letter. Things like ǻ or ệ are big no-no's.

Despite all this, there are diacritics that look perfect in some consonants: ç, č, ğ, ŕ, ś, ÿ, ž... Basically the c-cedilla and acute accents, carons and umlauts.

For vowels I prefer grave and acute accents (simple and double), breves and macrons, carons, diaeresis, rings, tildes and slashes.


And when I can use digraphs, I use them, but it really depends on the graphic aesthetics I want for the conlang.

[Thry]

Totally agree this one is ĥorrible; probably the main cause underlying Esperanto's defeat xDDDD.

[Pole, the]

Totally agree this one is ĥorrible; probably the main cause underlying Esperanto's defeat xDDDD.

You mean the diacritics or the digraphs? Because Esperanto has either.

[Thry]

Diacritics, especially ĥ

(inb4 _x, _h, no longer used, ...)

[2+3 clusivity]

I guess you could argue that the featural-alphabet like features of IPA (i.e. the retroflex series) are "diacritics." I don't think too many people complain about those relatively simple symbols--perhaps nor do many people complain about cedillas or tildes over vowels or <n>.

I guess diacritics and di/tri/etc-graphs are parallel solutions to the limited toolbox of the Latin alphabet. I think that the choice of one over the other really depends on how rich your vowel system is or how many secondary/co-articulations are present. The more complex things get, the more likely you are to use diacritics or diacritics combined with whatever-graphs; if you only have one series of secondary feature--aspiration let's say--then digraphs will be easier.

I think it is also interesting how aghast people are to a simple third choice, importing letters out of other alphabets--Greek, Cyrillic, etc. It can be tastefully done: http://commons.wikimedia.org/wiki/Categ ... vich_Uslar.

[din]

My conlang only has one diacritic, the macron or circumflex to mark a long vowel (you can use either, as long as you use the same diacritic consistently). It doesn't have all that many long vowels, so there'll never be many diacritics in one sentence.

I also much prefer digraphs, but what I like even more are languages with a relatively small amount of phonemes. My conlang Tormiott has 4 vowel phonemes and 19 consonant phonemes (of which 3 are basically long counterparts of short stops, which no longer actually contrast anyway, so we might as well put the number at 16).

Though for example /a/ can be [a, ɛ] in stressed syllables or [ɐ, ə] in unstressed syllables, and then there are things like /ow/ and /iw/ which become [u:] and [y, ʏ] respectively. I personally think it's a lot more interesting to have a bit of allophony than having 85 phonemes that can occur in all possible environments, which for that reason all need their own symbol.

[Cedh]

I think it is also interesting how aghast people are to a simple third choice, importing letters out of other alphabets--Greek, Cyrillic, etc.

I actually tend to prefer importing letters from other alphabets, especially from IPA, but that's only useful for certain phonemes in certain phonological systems. Polygraphs tend to work best if an alternate analysis as complex segments (e.g. prenasalised, aspirated or glottalised stops, labialisation, palatalisation...), while special characters and diacritics tend to work better for less systematic stuff like additional POAs or individual phonemes that don't have a straightforward equivalent in the Latin alphabet. For vowels, I prefer IPA symbols or diacritics wherever possible. If I have to use diacritics on both vowels and consonants, I generally try to put one set above the base letter and the other set below it (see the Cəssın, Tmaśareʔ and Omari examples).

• Buruya Nzaysa uses the polygraphs mp mv nt nts nz ŋk ŋkw kw (some of which contain the special character ŋ) and the special characters ñ ’ ɛ ə ɔ. Diacritics only occur in the form of acute accents on vowels in irregularly stressed syllables.
• Ndok Aisô uses the polygraphs mp mb nt nts nd ng ngg ts p' t' k' (the prenasalised stops can be analysed as clusters though, and I would probably replace the apostrophes with h if the aesthetic of the language was still changeable), and the diacritics ê ô.
• Cəssın uses the polygraphs ny ry ly cy zy, the diacritics ş ç å ä ö ü, and the special characters ə ı.
• Tmaśareʔ uses the diacritics ć ś ń ą ę ǫ and the special character ʔ.
• Doayâu uses the polygraphs mb nd ng and the diacritics á é í ó ú à è ì ò ù â ê î ô û (designating tone; the circumflexes only appear on diphthongs and the acute and grave only on monophthongs).
• Tsemehkiooni uses the polygraphs ts dz, the diacritics č š, and the special characters θ ʔ. It is also the only one of my languages which uses double vowel graphemes to indicate length. And an interesting detail: The language uses voicing on word-final obstruent graphemes to indicate the stress pattern of a word - if the final vowel receives stress, a voiced grapheme will be used.
• Omari uses the diacritics č š ą ę į ǫ ų.
• Hkətl’ohnim uses the polygraphs p’ t’ ts ts’ tl tl’ č’ k’ kʷ’ q’ qʷ’ ny (some of which contain the special character ʷ), the diacritics ł č š ħ ā ē ī ō ū (I guess for ł ħ it's debatable whether these should count as diacritics or special characters), and the special character ə.

In the end it does come down to the specific aesthetics and "feel" of the language, as Naeliums said.

[Matrix]

For me, it depends.

Maja romanization has no diacritics or special characters. The only digraphs that appear in the romanization are th for [θ] and sc for [ʃ]. Sc is the only one of these also represented by a digraph in the Maja Alphabet.

Standard Nahakhontl romanization has a bunch of letters with diacritcs: íôéúžĥšč. However, all of these can be represented digraphically for ease of typing. Aside from í and ĥ, they are the base letter with an h after it. ĥ becomes kh, and í becomes simply i, where standard i becomes digraphical ih. Also, the standard special characters þ and ð become digraphical th and dh. Oh yes, and in both versions, [t͡ɬ] is represented by the digraph tl. Aside from tl which is represented digraphically, all these have single letters in the Nahakhontl Alphabet.

Salenzian romanization also has no diacritics or special characters. The digraphs that appear are th, sh, zh, ch, ty, dy, sy, zy, wl. The -y digraphs are palatal allophones of the alveolars that I only distinguish in the romanization for my own convenience. Everything else there is phonemic and is represented by a single letter in the Salenzian Alphabet.

Ishdes romanization has no diacritics or special characters, and only one digraph, sh.

Budisua romanization exclusively uses digraphs instead of diacritics or special characters. They are as follows: si, su, zu, li, lu, ni, nu, vr.

Ta-lrm-zrq'-k'r has a bunch of special characters and diacritics which also have polygraphic alternatives for ease of typing. Standard ŋ, ɲ, ⱪ, ɠ, ɋ, ç, ᶍ, ƀ, ᶅ, ɭ. Polygraphical ng, nc, kh, gh, qh, ch, xh, bb, lh, lxh. There are also the standard digraphs tᶅ, dɭ represented polygraphically as tlh, dlxh.

Epœcuɒtœ romanization has special characters for some vowels: ø, ɶ, ɒ. It uses polygraphs for consonants: nj, ng, mw, nw, njw, ngw, mh, nh, njh, ngh, pw, tw, cw, kw, pj, tj, kj, ph, th, ch, kh, pq, tq, cq, kq, sw, sj, sq.

Dolkharezan and Ancient both have no diacritics or polygraphs.

My newest language, Ĩŋguri, has special characters, diacritics, and digraphs in its romanization: ŋ ṗ ṫ ċ ẋ ǰ ơ ā ō ã ẽ ĩ ỹ õ ỡ ũ ǟ ȫ, palatalization is -j, and labialization is -w.

[sangi39]

I tend to go for diacritics in romanizations over digraphs, but I try to limit myself to the more common ones, e.g. haceks, acutes, umlauts, graves, etc. I will use digraphs where the syllable structure allows it, but in the romanization challenge thread I most often use diacritics to achieve a 1-to-1 correspondence and then see if any unambiguously digraphs could be used instead if it ends up being to diacritic-heavy. If there's no syllable structure or sample text given, I'll go for the 1-to-1 correspondence as a default.

[clawgrip]

For any conlang I make a usually will create a script to go with it, so I treat the Romanization as a simple tool that I try to make as simple and straightforward as possible. As such, I generally just start with English as a base and will employ digraphs such as <ch> and <sh> etc. as needed. With vowels, though, it depends on the language. If I can get by without digraphs I often will, but when there are many vowels, I may use diacritics or vowel digraphs depending on the overall vowel system of the language.

[Nortaneous]

I think digraphs are more suited to languages with shorter words and diacritics are more suited to languages with longer words. Cherun and Ngmwragh both use only digraphs. (Although I'll probably introduce a tilde into Ngmwragh for the labial-velars and palatal nasal.) Kannow and Miar use only diacritics, as does Hanheliubl.

Part of it is that I don't like Ch digraphs for things other than aspirates -- Hanheliubl would have a lot of <sh zh> if I used digraphs. But part of it is readability. <ğbə̣̄m̃> is more confusing than <ghbeurngm> to me, and I can read <cëʔëdgyagwënṟt> much better than <ceiʔeidgyagweinrht>. I do what I can work best with.

[Imralu]

The only diacritic I use for Ahu is the acute accent to show stress where it's not on the first vowel. Other than that, there are digraphs but they are not arbitrary. Affricates are written as digraphs comprised of their component parts (ts, tc, dz, dx). They're only considered phonemic because they are allowed in syllable onsets while clusters aren't. Then there are the "strong consonants" which are pronounced as ejectives initially or after a consonant but generally after a vowel as a geminate (pp, tt, kk).

I have an aesthetics question: What looks better for ejective/geminate affricates? Tts and ttc or tss and tcc? I think the firmer pair make more sense but the latter look better.

Sorry about the lack of formatting. It's hard on my phone.

[din]

I think the former makes more sense and looks better.

[Thry]

I agree with Imralu, having to choose between one or the other. tss > tts.
Well, actually it depends, <tcc> is uncanny.

[Herr Dunkel]

I use <sh zh> for /ʃ ʒ/, but <ā ē ī ō ū ǣ> for /aː ɛː iː ɔː uː ɞː/ so it's balanced I think.

[Burke]

Honestly, for Romanization I prefer using digraphs over using diacritics. I have no really strong reason for doing it. I simply can normally work it out such that it is less Key strokes that using diacritics. So laziness is pretty much my reason. I have used diacritics before, though. I try to keep them off of consonants because they tend to be more of a pain than they are worth, and when I need them for vowel distinctions, I try to put it on the least commonly occuring things.

I guess I could switch to Spanish or Portuguese input, since I don't think twice when I use diacritics when typing in Greek since it is typically so easy to do.

As long as I can get the romanization/orthography to be consistent and not a hot mess by these means, I tend to stick by them.

[Nortaneous]

hmm

• Kannow uses the digraphs pʼ tʼ sʼ lʼ rʼ cʼ kʼ qʼ aa ee oo ii yy uu, the diacritics ŗ ņ į ł ř ň + acute, grave, circumflex on all vowels, and the special character ʼ.
• Miar uses the digraphs ph th sh lh c̣h ch kh qh, the diacritics c̣ ṇ ṛ ṉ ṟ ñ ä ë ï ö ü, and the special characters ʔ ŋ.
• Enzielu (when I get around to it) uses the polygraphs zv sv xv tz ts tx tzv tsv txv aa ee ii uu and the diacritic ĕ in the standard orthography, and the diacritics ż ṡ ẋ żv ṡv ẋv ŏ oo in dialectal orthographies.
• Insular Kett uses the digraphs ng tw and the diacritics ḩ á é.
• Continental Kett uses the digraph ng and the diacritics ṯ ḵ ö ü á é í ó ú ő ű.
• Proto-Hathic uses the digraphs ph p̃h th ch kh hw tl ʘh |h !h ǁh ǂh mʘ n| n! nǁ nǂ bʘ d| d! dǁ jǂ aa ɛɛ əə ɔɔ ee oo ii ɨɨ uu, the diacritics p̃ b̃ ɓ̃ m̃ ñ ḷ, and the special characters ʔ ɓ ɗ ʄ ŋ ɕ ħ ʎ ɣ ʘ | ! ǁ ǂ ɛ ə ɔ ɨ.
• Cherun uses the polygraphs ph bh th kh tl kl ll lly my ny ar ur or aa ae ee ii oo uu aar er eur oor ai ei ea eu ua ie ue and no diacritics or special characters.
• Ngmwragh uses the polygraphs kp ph th ch kh kph |h ǁh !h ǂh n| nǁ n! nǂ gh ny ng ngm ar er ur or aa ee eu oo ii uu aar eer eur oor, no diacritics, and the special characters ħ | ! ǁ ǂ.
• Arve uses practically innumerable polygraphs, of which ck tj kj tz lt ld lz rz zr kr gr rg jj dg dj gj sl ls rs sr rst str nts st sk ks stj js hj sg gs ch chs sch kl lk hv vh chv gv vg sv rch rk nk nch nv hd chd rv mb vr rj lj gj nt tr ntr nz ns dr rk rg ng lg rn gn el ol ug gu uch äh ei öy ou are currently attested, and the diacritics ä ö ü.
• Hathe (when I get around to it) uses the digraphs th ii and the diacritics ā ē ō ī ū.
• Hanheliubl uses the digraphs mh nh ñh ŋh lh ai au iu, the diacritics ṭ ṣ ẓ ñ ẹ, and the special characters ŋ ɯ :.

[WeepingElf]

In my romanization of Old Albic, I use all three ways of extending the Latin alphabet, each in modest amounts:

* Digraphs: ph th ch ng for /ɸ θ x ŋ/.
* Diacritics: Acute and circumflex for long vowels with thrusting and slipping tone, respectively.
* Special letters: ø for /ø/.