Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[RedKing]

Okay, a little more lexical analysis seems to indicate 18 consonants: b, c, d, f, g, k, l, m, n, r, s, t, v, z, st, sh, th, ng. There are the occasional irregular consonantal clusters like kr (ex. sákrith, blaze), but these appear to be illusionary clusters caused by two adjoining CVC syllables (eg. sák + rith rather than sá + krith).

These CVC syllables are very common, which means an abjad (similar to Arabic) could work. In the above example, we'd write s + k + r + th, with diacritics to indicate the two vowels (and the lack of a diacritic between the K and R indicating no vowel). There are at least 5 accent marks used on vowels (á, ŕ, â, ä, ĺ, but not all vowels receive all accents. So I'm not entirely sure of the total range of vowels. At least 12, possibly as many as 30. Rather than have seperate glyphs for the accented vowels, I'd probably use an additional diacritic such as top/bottom marks to indicate accented vowels. 

The other thought was to have a set of characters for CV clusters (sa, se, si, so, su) and another set for VC clusters (ak, ek, ik, ok, uk) and a mark to indicate when they're being combined (sa + indicator + ak = sak; sa + ak without an indicator = sa'ak). Definitely more characters to create, but potentially much easier to use. Might even be quasi-alphabetic if the vowels are indicated mostly with diacritics. CVC clusters would wind up as a two glyphs with a single vowel "combiner".  Maybe a line connecting the two consonant glyphs, with the angle/direction of the line indicate the basic vowel:

- = a
\ = e
/ = i
V = o
^ = u

or something like that. Accents could then be denoted by dots above/below the vowel marker.


Given that there's no inherent clusters of meaning, dropping the vowels altogether is unlikely. (in Semitic languages like Arabic and Hebrew, related words often have the same consonant structure, so in some cases the vowel marks are dropped completely and a text only indicates the consonants. Vowels are subconsciously added by the reader based on context (if you don't believe this is possible, jst tk _ lk t ths sntnc nd yll s wht _ mn) 

[RedKing]

Okay, had a brainstorm while at lunch.

Following the above idea, if we use blocky, geometric characters for the 18 consonants and "joining" marks for the vowels...it actually becomes feasible to use the characters within DF itself by mining them out or building them. It's not dissimilar from building a row of workshops with short passageways in between them. Because the different workshops have different "entry points", sometimes you have to slant the passages.

The fun part is that this means you could literally carve out a section of text in the fortress layers themselves. You could build giant letters out of marble blocks or what have you that spell out the name of your fortress! 

You could have a megaproject to inscribe some phrase in a huge box around your actual fortress!

The trick is figuring out what scale to make the characters. 3x3 is far too small, 5x5 probably is too. Might be able to come up with a decent set of 18 distinct characters in a 7x7 block, but I'm not sure. It needs to be an odd number so you get a pure symmetry for making diagonals and center marks.

With that, I think the five base vowel markings would be a high straight bar, a middle straight bar, a low straight bar, an ascending slash, and a descending slash. Probably smaller than the block consonants (if the consonants are 7x7, maybe the straight lines are 3 or 5 blocks long? The slashes would need to be 7 long if you want to preserve a 45-degree angle; or you compress the angle of the slashes and make them 3 or 5 blocks wide).

Thoughts?

[Vattic]

Being able to use it in game would be kind of neat, not sure if it's something I'd do myself, though. Obviously smaller would be better which makes it unfortunate that, with such a limited colour palette in game, we can't do any anti-aliasing. I think 5x5 would be doable at a push unless you wanted to have a continuous line through each word.

EDIT: One nice side-effect of all this is that even though Toady (or ThreeToe...not really sure which one is more involved with making dwarfy words) aren't really using any kind of ruleset to generate new words, there *is* an emergent logic to the vocabulary. For instance, as far as I can tell, there are no "w" sounds in Udosnol. This would allow us to create rules for use in a basic "word generator" program (of which I have several) and create additional vocabulary if so desired, that would still "sound" right.

From what I remember Toady wrote some software to generate the words that accepts rules. He then goes over them to remove anything that is too close to real words and tweaks things generally.

[Skyrunner]

S\\ms pr//ty sV^nd.
... and I know you aren't supposed to use them like in english but ... xD

Would be reasonable to actually learn, I think.

[Vattic]

Just as an example I've made some runes in 5x5 (300% zoom).



Obviously we'd be limited to something visually similar as variations on the theme is all you can really fit. Also making letters out of straight lines like this also limits some.

Edit to add: If we start small we can always make a larger more ornate version.

[MonkeyHead]

I could see dwaves using some kind of cuniform type system, where the strokes of a "letter" could be laid with a single chisel strike for utilitarian reasons.

[RedKing]

Well one thing I need to investigate is that (as I'm picturing it in my head), each consonant would technically have six "contact points" -- three on each side -- for the vowels to connect to. Because yeah, I was kind of picturing it as a script where a whole word would be one unbroken line. Since the vowels are going to be straight lines (at different angles), the diacritics could be simple short lines at a 90-degree angle to the vowel. (note to self: I need to draw and scan in some of what I'm seeing in my head, to help illustrate the point).

Obviously that's an even bigger constraint if all the consonants need to have "connections" at six cardinal points. BUT....early lexical analysis indicates that not every consonant will need all six. For instance, I've yet to encounter a z followed by an o. And c is always the first part of a CV cluster, and never appears in a VC cluster or as a final consonant, so it wouldn't need any "connections" on the left side (assuming it reads left-to-right). It's almost like working out which workshops need which door locations.

I'll probably work on this some over the weekend, building up a matrix of which pairings are present and which aren't, and from that work out which connection points are for which vowels, and then from there that'll give us our constraints on the shapes of the consonants. Really getting excited about the potential here. Even if it's purely an artistic flourish, I'm sure there are some in the upper forums who would enjoy it.

[SirAaronIII]

Posting to watch. I don't have anything productive to say, just agreement with previous expressions.

[RedKing]

Okay, looks like c, f and g are "tailless" consonants, in that they're never a final consonant, and never appear before another consonant, only vowels. So that opens up a few options in glyph design for those.

Looks like all 18 consonants can appear in front of every vowel, and 15 of the 18 (the three exceptions above) appear behind every vowel in a VC cluster somewhere. So all in all, that didn't change much. I'm kind of leaning back towards an abugida-type script, where the consonants are assumed to have an inherent vowel after them, and we can use diacritics to denote the vowel (or lack thereof). The "continuous line" idea is going to be tricky to pull off because not all words are a constant CVCVCV type pattern. Good example is mabdug, "ale". The form is CVCCVC. Connecting the B + D is possible but potentially irregular.

If we do it as an abugida, it becomes (assuming the inherent vowel is "a"): M + B + (vowelless diacritic) + D + ("u" diacritic) + G + (vowelless diacritic) = 7 distinct parts

If we go back to the idea of focusing mainly on CVC constructions (using a syllabary), it would be: MA + BA+ (vowelless diacritic) + DU + GA+ (vowelless diacritic) = 6 distinct parts


Although now as I'm thinking about it, I'm noticing that Udosnol words ALWAYS and without exception, end in a consonant. So the last diacritic could be dropped, as readers would inherently know to drop the vowel. So the preceding options would be 6 and 5 parts, respectively. Now we're getting somewhere, as it's approaching the same or better utility as English. xD

Another consideration I should look at is how common are certain final consonants vs. the same consonants as initial CVs and the vowels used. I'm thinking of Japanese, where "su" virtually drops the u-sound if it's anywhere other than the initial syllable. So for instance, sukiyaki and sumitomo clearly pronounce the u, but desu is pronounced "des", gozaimasu is pronounced "gozaimas", etc. Same thing for shita: if it's anywhere other than the beginning, the intermediate i-sound is so clipped, it sounds like "shta".


ANOTHER idea: What about a reverse alphasyllabary? That is, instead of ba/be/bi/bo/bu, we have ab/eb/ib/ob/ub? That takes cares of never having to deal with final consonants again. Beginning consonants in a CVC cluster could just use the a+consonant glyph and a diacritic to indicate the vowel is dropped. So then mabdug would read as AM + (vowelless diacritic) + AB + AD + (vowelless diacritic) + UG = 6 parts. Utility isn't really any better in this case, but it's potentially got a lot less irregularity to it, at the expense of using a LOT of initial diacritics.

[FearfulJesuit]

Have you considered something like Inuktitut? Inuktitut is a syllabary, but final consonants, like q and k, are diacritics formed from the qa/ka forms:

ᑐᑐ 'tutu' = "tangles in hair"
ᑲ 'ka'
ᑐᒃᑐ 'tuktu' = "caribou"
ᑐᑐᒃ 'tutuk' = "dirt"

[Thief^]

Personally I like the abugida stuff you've been doing so far, especially if word ends are assumed to have no vowel (and other consonants assumed to have a vowel). You could do the "vowel-less" diacritic as being a small gap between letters, and the word gap be a letter-sized space, perhaps. Both would be vowel-less.

[RedKing]

I suppose another option, if we're doing vowels as various straight lines, is that the vowelless marker could just be a very short line connecting the two consonants. I need to get down to making a few sample glyphs so I can play around with spacing and seeing how that'd look.

[RedKing]

Got an initial set of glyphs cooked up. Not real thrilled with them because they're not terribly original or memorable (and very, very blocky and repetitive).

I'll try to jot up something and post it for comments later tonight.

[RedKing]

Just a quick bump to note that I haven't abandoned this, just been interrupted by RL.

[DeKaFu]

Not much to add, but this is really cool. I'm looking forward to seeing what comes out of this.