Page 1 of 4 1234 LastLast
Results 1 to 10 of 31

Thread: Conlang construction philosophy? :)

  1. #1
    Guild Expert Jalyha's Avatar
    Join Date
    Jan 2014
    Out of my Mind!
    Blog Entries

    Default Conlang construction philosophy? :)

    Hi there!

    Maybe because I speak 18 languages (only English is spoken by any other person on the planet), and maybe because it's come up in a lot of conversations recently, I ended up doing a 16 page rant on conlangs on another forum.

    No one there was even vaguely interested in anything about conlangs (it's more of a vocab/etymology/we-love-words type of site, but anyway...) I'm still worked up, and curious, and I figured that I'd share my own thoughts/methods here, and beg everyone to share their own thoughts as well.

    And don't worry, it won't be 16 pages... or a rant

    Anyway... If you look at any "alphabet" ... or whatever basis for written language exists, you will find that there are very few actually *different* shapes/symbols used.

    The current "English" or "Roman" alphabet, original the latin alphabet, was and is, in both forms, straight lines, half circles, and circles. They are PRIMARILY straight lines. Those straight lines can be combined in any number of ways. Here's a sample of less than 1/3 of the symbols you can make with *no more than 3 straight lines* (the letters E, M, and W all have 4 lines, btw).

    There are 42 symbols - that's more than the number of letters in our current alphabet (though some languages have many more) (26) AND 9 digits, with 7 more to indicate 10s, hundreds, thousands....

    Click image for larger version. 

Name:	lines.jpg 
Views:	816 
Size:	74.2 KB 
ID:	60261

    Remember, this is just a small sample of the symbols you can make with 1, 2, or 3 straight lines. Yet NINE of our letters are approximated in just that sample. More than 1/4. by curving those sample lines only slightly, you cn make 11 of the letters in an arabic alphabet, and 12 *words* in chinese.

    A lot of conlangs are simple reworkings of a current language. Many people will sneer at these languages, because they prefer a language with grammar of its' own.

    Many conlangs are based off an existing alphabet (with new grammar rules) without knowing why those alphabets are the way they are. Some people turn their noses up at these, because there are a few gems of conlang that have their own alphabet.

    That's fine if the alphabet makes sense in the setting.

    If your people came (originally, no matter how far back) from Earth, then yes, their languages/alphabets can be familiar. And you can say those languages developed however you wish.

    On another world, however, languages will develop differently.

    Our written languages start with straight lines because we, as a world, in many different cultures, all started writing before we had anything to write *on*.

    Try writing DANGER! in Elvish. Now carve it in the wall of a cave with a sharp rock. No? How about on a piece of wood with a stone knife? Can't do it? Can you write it with a stick in the dirt?

    Okay, okay, elves are more advanced than we are, and their language had developed past ours before we hoomans ever existed. Got it.

    At least there's a reason for it.

    Where was my point? Oh yeah.

    I've always figured that the best way to create a language is to begin at the beginning. When did my people first start communicating with words other people could understand? When/how/why did they start writing that way? Did they begin with berry-stain painted pictures that were later simplified? Did they ALWAYS use letters? Do they still use pictures?

    Then I consider (based on when they started communicating) what was most important to them at that time. It's almost always food, water, shelter, and danger. Those are your real root words.

    Later I add words for family, for "mine" and "yours". For good, bad, up, down, far, near.... Usually, I let the rest of the language flow from there. Other words develop more naturally this way. Words for my family can grow from words for home - which is basically *my* shelter. That food is dangerous, but this one is just bad. Tastes bad. How do I tell my neighbor not to eat it?

    My final inspiration, when struggling to create (or begin) a new language, is to go to a neighbor's house, lay on the floor, and talk to a baby.

    What sounds do human beings make most naturally?

    Or preschoolers- what sounds are hardest for them to learn?

    Why else are there, in 100 different languages only a few ways of saying "Mom" which don't sound similar?

    That's one word I always develop from baby sounds.

    So I'm curious. How do other (less obsessive/perfectionist/semi-psychotic) people develop their conlangs. What kits do you use, what guides do you follow, or how do you create your own?

    (Eh, maybe it was a bit rant-ish, after all... sorry )

  2. #2
    Guild Adept
    Join Date
    Jul 2011
    Rome, Italy - New York, USA


    I have always (since reading Tolkien, of course) been interested in creating languages; but, not being a linguist, nor having a particularly strong work ethics, I haven't ever been able to proceed very far. The most developed language I have created only has an incomplete simple grammar and ~240 words.
    Not having any training as a linguist, I did not proceed in any systematic way to create it, and I don't even know the right words to describe what I've done (not in English, to be sure). However, back in high school (~15 years ago) I had created an alphabet for a previous version of this language, and I had even thought about the way in which letters could have developed as to be easy written with simple instruments.
    Snippets of the language are visible in many names on my maps, though sometimes finding names I like means breaking some rules (which may be ok, as real languages change and evolve... or may be just another sign that I am not a linguist by any means).

  3. #3
    Guild Expert Jalyha's Avatar
    Join Date
    Jan 2014
    Out of my Mind!
    Blog Entries


    I'm no linguist either - (though I'd like to be, lol).

    "thought about the way in which letters could have developed as to be easy written with simple instruments." That's all I use really, simple common sense, and looking around at other languages.

    240 words, an alphabet, and even a sense of grammar is lovely. (Especially since you were in high school!) Languages don't develop overnight. A complete language takes generations, and never stops evolving.

    Even if a few of your names break some rules, well, there are exceptions in any language.

    I think the *biggest* pitfall for writers is using too *much* of their language in their books - we're writing for people to read, so it must be understandable, lol.

    I'm very interested in how people begin and what they use for inspiration... I've seen people name everything - write whole speeches in a conlang... and then make up what it means and why (which seems difficult to me), and I always start at the beginning...

    I also like reading them and seeing if I can puzzle out the words/rules, lol.

    What do you like about con-languages?

  4. #4


    I'm glad to have stumbled on this thread and I'll be interested to see where it goes!

    I've started many conlangs but never gotten very far. Since my last serious attempt, I've developed the ability to read in five languages other than English and I've studied a few others, so I'm hoping that I'll be a lot more informed on my next serious attempt. For the present, I'm going to base simple words in each conlang on words from a designated language that I've studied, changing them to be unrecognizable but still conforming to the pronunciation rules of that language. Although I suppose it might be more interesting to adapt words from one language using the protocols of another. Hmm...

    So I don't have a system yet, but that's where I'm going to start. As far as creating alphabets/syllabaries, I intend to do it, but ultimately I expect that everything will be written in English. It will be nice to have the alphabet in mind, but as long as the sounds of the language are firmly established, it's really just ornamentation. That's from the point of view of someone who wants for many people to enjoy his work and world--I'm sure it's different for different purposes.

  5. #5
    Guild Journeyer rgcalsaverini's Avatar
    Join Date
    Jan 2014


    Great! I found someone as obsessed as I am with conlangs (although clearly more skilled at it)!

    Could you post the link for your 16 page post about it? I would love to read it.

    I'm an absolute amateur, and to be honest I never heard of the term 'conlang' before this post

    Please share one of you languages with us!

    To assist me on creating languages, I wrote a small piece of software that devises word-formation rules from a set of pre-established words, and then using those rules it gives me a list of original words. For example, I just fed the program with the following list:

    compass, cartographer, guild, map, draw, conlang, language, example, banana, bowel, barnacle, garnish, negative, nile, novel, gorilla, love, amiable, intrusive, meridian, bent, bongo, cilatro, positive, inclination, movable.

    And then asked for 30 words that follow the same basic set of rules, and it gave me the following:

    gamono, morine, nivaborn, govenile, barn, powegove, banamele, bowern, merone, ilivaw, blovern, minto, coname, meldrua, bernge, cilelass, gasshele, nisillel, nion, gruaw, ivelero, erapan, gativap, gasshe, cowertia, guilapos, banit, garusshe, lerampo, bome

    That way I avoid being biased by the languages that I do speak and helps to make sure that all the words follow somewhat the same pattern.

    I'll show a conlang that I'm currently working on, Humir:

    Men of humir (Humiren):
    Name:  humiren.png
Views: 614
Size:  1.7 KB
    I used a list of 50 thousand words in norwegian (25%), swedish(25%), danish(25%), german (10%), english (10%) and spanish (5%) to generate all my words, I also adapted the final rules a bit by hand. To make it plausible I first start with a basic set of words then I form others deriving from those: Alüne (hard, strong) + Inäk (soil) = Alüniak (stone)

    Then I made a few decisions about grammar. Regarding declension, I decided that Humir, being quite primitive, would not inflect by gender, number nor case:

    Declension - gender: Change nouns, adjectives and etc to show gender:

    English (not inflected):
    The ugly frient (male)
    The pretty frient (female)

    Portuguese (inflected):
    O amigo feio (male)
    A amiga bonita (female)

    Humir (not inflected):
    Malor igep
    Rikmalor igep

    Name:  gender.png
Views: 635
Size:  4.4 KB

    Declension - number: Show quantity:
    English (inflected):
    One rock. Two rocks

    Humir (not inflected):
    Ige aluniak. Tje aluniak.

    Click image for larger version. 

Name:	number.png 
Views:	63 
Size:	4.5 KB 
ID:	60279

    By having so little inflection, the word order becomes crucial for the understanding of the phrase, as is the case with most ancient languages (i think). Take this phrase from the Brazilian national anthem: "The quiet margins of the ipiranga heard the echoing yell of an heroic people", in a more inflected language, such as portuguese, It could be written as:
    Ouviram do ipiranga as margens plácidas, de um povo heróico um brado retumbante -> Heard of the Ipiranga quiet margins of an heroic people an echoing yell"

    Also regarding inflection of verbs (conjugation) we have tense, mood, voice, aspect, person, number, gender and case. Likewise i decided that Humir verbs will not be inflected at all, and if the phrase is ambiguous the person, number, tense etc is explicitly declared before the verb, for example:

    John is loved by Mary -> John *he one passive love* Mary (This is just for argument sake, the number and person you can extract from the phrase, and it would be written just as *passive love* )

    Click image for larger version. 

Name:	verbs.png 
Views:	290 
Size:	32.4 KB 
ID:	60280

    The alphabet I created in the same manner as you guys do. Unlike another language I created, that derived the alphabet from their drawing habbits, the humir developed a necessity for quick and simple writing based on straight lines, since they were not very big on drawing. Here is an example:

    Here lies Lord Khoven, protector of the Green Lake Village, son of Swol-Overu, grandson of Iderak-Demku, born in 321 and murdered in 357 of the second age.

    Gdak figör ige askjuetese ral (Khoven) foski oda (Halikje Some Tagra), igefüp oda (Swol Overu) ge tjefüp oda (Iderak Demku). Figör ige üllve timitüs pjo 321 ge figör ige üllve kedetri pjo 357 pjo pravet tje.

    Click image for larger version. 

Name:	phrase.png 
Views:	109 
Size:	32.2 KB 
ID:	60281

    The ovals around some words shows that it is a proper noun, since most names have clear meaning (Strong Bear, Fast Horse...)

    I'll stop here because this post is already longer than I wanted!

  6. #6
    Guild Expert Jalyha's Avatar
    Join Date
    Jan 2014
    Out of my Mind!
    Blog Entries


    "although clearly more skilled at it" --- not really... I haven't had any classes or done much real research... as I said, I usually start from basic common-sense. I don't even know the right terms for some of the things I use/do... lol.

    As for your first request, I have to decline. After one incident of identity theft, and another instance of an internet stalker (who came to my home!) I do not share personal information anywhere, nor do I intermingle my websites. My forums must remain eternally separate. I don't mean to offend anyone, I'm sure no one here is like that, but we all have our boundaries... I'm sure you will understand.

    I might, however, be willing to rant at you for 16 pages in a pm

    Rikmalor Humir ... ? = Humir is beautiful? Or I've missed a step?

    Either way, it's lovely! Did you actually transcribe all 50k words?! That's amazing. Most people don't get nearly so far! I'd love to get my hands on that software. I, unfortunately, must do everything by hand.

    I tend to be lazy and avoid gender declension, altogether. I'm no linguist, despite my passion, and I always worry about messing it up. Since I don't speak any actual foreign languages, just my made-up ones, and English is my 1st language, I am a bit limited by my own ignorance as far as *spoken* languages.

    I can usually understand at least the rudiments of written languages, however, although I'm better with those of latin origin.

    One thing I'm very careful of, when initially compiling my root-word lists, is only pulling from languages that have a common base.

    You said you "used a list of 50 thousand words in norwegian (25%), swedish(25%), danish(25%), german (10%), english (10%) and spanish (5%)" I would have excluded the English and Spanish words for this stage.

    Languages do tend to pull from other languages/cultures, taking a word here and there when they have no other term for it. However - that doesn't usually happen till the language itself is more fully developed.

    I'd be interested in knowing what words you transcribed to start.

    I start much smaller, of course, doing everything by hand. I start with:
    water food
    Woman Man
    Child Infant
    Tree Mine
    Not mine Want
    Need Danger
    Stop Ground
    Tree-food (fruit) Ground-food (veggies)
    Animal-food (meat) Weapon
    Hot Cold
    Up Down
    Far Near
    Tool Good
    Bad Fight/Battle/Hunt
    Come Go
    (Various numbers)

    Once I have those words down, I work on a basic layout of how the words fit together. (Grammar, but beyond basic into almost non existant) Then I add words for:

    also a/the/this
    covering (hat/hair) top
    bottom unseen
    tame predator
    relating to a person nature-related
    objects feelings
    philosophy related to learning
    time not/against
    seperate from ability
    sour/bad good/sweet
    now! finished
    (various colors) love/fondness/desire
    big world/not of the world
    round/smooth rough/sharp edged
    before self
    with reduce/undo
    between/among past
    same different
    bad many
    in favor of again
    false/untrue big big!
    better Bush
    Rock Grass
    Pigs Cows
    Chickens Goats
    Sheep dogs
    bears fur
    work home
    children spouse
    parent peace/end fighting/battle
    ruler/leader house
    table seat
    bed horse
    bird wash/clean
    clothing (maybe) shoes
    weight good/well
    purse/pouch yield
    conquer center
    speak hear
    see touch/feel
    taste herbs
    material injury
    death hard/difficult
    beauty faith
    old new
    sun moon
    (various gods/goddesses/stuff)

    This usually takes quite some time AND gives me enough to start writing whole sentences/stories in the language. Once I'm good enough at that where I don't have to cheat/look, I write out all the grammar rules and start adding to the dictionary.

    When you actually speak/write the language, it's easier to tell when a word doesn't fit.

    Currently, I'm working on a (still-unnamed) language for the people of Naos (on the planet Naos - cause of course their land is the whole world right?) I'm still at the caveman stage, though, so I only have my basic word list, alphabet (or syllabary, really), and a few cobbled-together grammar rules. The entire language is made up of 88 letters, (or syllables).

    na- used to emphasize something. It can mean "big" (when used with another "a" sound consonan) Na-ra, Na-ka, Na-va) or "right away" when combined with a "y" sound which indicates action.(Nayotu - literally, "Right now - come/go - labor" which means, in essence, "Get to work!!"

    central syllables are broad applications to a thing, surrounding syllables define it. "Va" is, by itself, a ruler or leader of some type. Na-va, is a big ruler - king or emperor, depending on the 3rd syllable.

    ri is drawn as a sun&moon, and is only used in describing Gods, or abstract concepts.

    os -dirt/earth

    Naossatu subject or nominative the world (is, does, ...)
    Naostusv object or accusative (something affects) the world
    Naosstu vocative O world!
    Naossotu possessive or genitive the world’s
    Naossutu indirect object or dative (given, sold, etc.) to the world
    Naossytu ablative (something is done) by the world

    In learning to read/write, young children on Naos are taught with the syllables marked separated with a simple dash - or dot between them. This dash/dot is pronouced with a "j" sound, (which is not even noted in the "alphabet" and making the sound while speaking is considered offensive (to either the listener or the subject at hand), while a pause between each syllable is considered either sarcastic, or a way of talking down to someone. Similarly, writing with dashes/dots is only done as a means of educating someone, or insulting them, except at the end of a line, when it indicates a joke or jest.

    I don't have a scanner, and my attempts to snap a pic of my written alphabet have failed I do use unicode letters/symbols to effectively write in my own language online, but I haven't assigned them yet... However, the syllables I use for the language of Naos are:

    b ba bo bu by
    d da do du dy dv
    f fa fo fu fy
    g ga go gu gy gv
    gh tch qtl tvl
    h ha ho hu hy hv
    k ka ko ku ky
    l la lo lu ly lv
    m ma mo mu my
    n na no nu ny nv
    q qa qo qu qy
    s sa so su sy sv
    t ta to tu ty
    v va vo vu vy
    w wa wo wu wy
    y ya yo yu yv

    (all the isolated vowels (and y) make a long-vowel sound, the others make a soft-vowel sound. I haven't gotten around to typing them up yet.

    In my notebook, I've a dictionary, so far of about 986 words (those listed above, and a few, tentative "next level" words.

    Anyway... I don't want to bore you with any more of my language... unless you want more, in which case I would be happy to upload the written language once I kick my camera around... and update you, if you like, as I work.

    WOW I keep doing such long posts.. I really don't mean to either


    Nasytuva dytlv sutuma. Nayotu -

  7. #7
    Guild Journeyer rgcalsaverini's Avatar
    Join Date
    Jan 2014


    "Rikmalor Humir" Bravo! Due to the very low level of inflection adjectives will always precede the nouns.

    "Did you actually transcribe all 50k words?". God no! I barely have time to eat properly, that would take me ages. I download lists of thousands of words on many languages from this site. The software works like this: given a list of words, it registers the frequency of every letter to appear on the beginning, end and middle of a word, the relative vowel to consonant ratio and the frequency of every pair and trio of letters.
    Take this list of words:
    babaca, cababa, abaca, baba, bac

    This is the rule the software devises from that:

    b: beginning 61% middle 50% end 0%
    followed by:
    a: 100%
    b: 0%
    c: 0%

    ... and so on.

    Then the software saves that rule file, and I can load it to generate words that follow that pattern, with that rule I got: cacaba, baca, caca, bacababa, babaca... I like using it because it have no creativity at all to create words, and they all end up looking like the languages I speak. So when I told about the languages mixture that I used, I was talking about mixing the rules, so my language sounds nordic, but with a little bit of english and spanish to make it more exotic.

    I know VERY little about linguistics too, I'm pretty sure that what I've done is all wrong

    I like very much the way you develop your conlang! I tend to rush it and devise the words mostly as I need them. Your approach on the other hand starts with the words that are more likely to be used by a primitive people, and build it up from there, this way your conlang seems much more consistent. I'll take a few steps back on mine and try to evolve it more naturally!

    Although it does seem to take a lot of time, it seems well worth it, I like your conlang very much! Please keep me updated on its development and post the alphabet when you have a chance =)

  8. #8
    Guild Expert Jalyha's Avatar
    Join Date
    Jan 2014
    Out of my Mind!
    Blog Entries


    EEK~! I downloaded/opened one of those word list and my computer yelled at me

    The FIRST problem I see with the list I opened is... well, I opened a list of English words, and several of the words I noticed in a quick (30 second) eyeball-check of the list is that there are a LOT of "words" that are really obvious typos and various typos of the same word. Like "wkno" / "nkow" / "know"... three of the "words" on the list.

    Next (still on the English words list) there were words that are not English... or possibly even words :s I don't know if they are even words in other languages without going through to check.

    Finally, there are a lot of slang words/phrases. That's fine too... except that those types of phrases don't usually follow normal grammar rules.

    That's all fine if you go through and check each word, and it's a language you KNOW.

    Otherwise, with a system that relies on statistics, you might be *starting* with flawed data.

    I don't see anything that says where/how the lists were compiled. I'd almost believe it was an internet keyword search, similar to search engines like Google... but I don't know.

    Starting from scratch and going step by step, it can take a HUGE time investment, but honestly I think you might be better off even starting with word lists from an online dictionary or something instead of taking chances with incorrect lists.

    What software do you use to analyze the lists?

    The biggest problem (besides the fact that it takes forever) with my method is that it's a bit TOO consistent. In later stages, it's a struggle to come up with origins for "figure of speech" type phrases, slang terms, contractions... things like that.

    It quickly goes from the free-flowing method we discussed in the pm, to something extremely rigid.

    Perhaps I should give you my word lists around the half-way point and let your software figure out the hard parts

    I'll post the written alphabet as soon as my camera stops fighting me.

    I'd love to see yours as well!

  9. #9
    Guild Journeyer rgcalsaverini's Avatar
    Join Date
    Jan 2014


    Computers do tend to be bitchy sometimes

    Yeah, the lists quality are passable at best. Actually for Portuguese, English, Spanish and Italian words I was able to get high quality lists from my university's NLP Lab, but they only had those languages Those lists you saw are compiled from movie subtitles (apparently pretty crappy ones). And even though I tried to clean them a bit, I reckon that at least 10% of its words are just rubbish.

    What software do you use to analyze the lists?
    I wrote the software myself, I dont have a name for it yet But there is a lot of room for improvement. I have a sketch of a pretty cool way to find patterns on words using artificial neural networks, I just don't have the time to implement it right now.

    The biggest problem (besides the fact that it takes forever) with my method is that it's a bit TOO consistent. In later stages, it's a struggle to come up with origins for "figure of speech" type phrases, slang terms, contractions... things like that.
    I haven't really thought about that, after a time a rigid structure could make things harder. But then I guess that you could just let your language get a bit 'corrupted' over time as you evolve it, just like it happens on real languages over time.

    Perhaps I should give you my word lists around the half-way point and let your software figure out the hard parts
    I'll be happy to help. I'm not absolutely sure about the software's output quality as of now, I do have plans for improvement. It does serve me well tho.

    I'll post the written alphabet as soon as my camera stops fighting me.
    Hahaha I'll be waiting.

    I'd love to see yours as well!
    I used the opportunity to make the pronunciation cheat-sheet that I've been meaning to. It is still kinda incomplete, some letter combinations change the sound a bit:
    Click image for larger version. 

Name:	humir_beaba.png 
Views:	744 
Size:	117.5 KB 
ID:	60308

  10. #10


    I actually have a half-constructed ancient language of my own. It is a syllabary and I haven't gone into any of the more advanced grammar and such. I just use it for individual words. The alphabet itself is very curved in style.

    Which reminds me - I once read that the shapes of the letters tended to follow the medium on which they were most often used. For example: Roman and Norse script was most often (originally anyway) carved onto stones which is why it is made of straight lines and harsh angles. Arabic script and the like was more often written with brush on papyrus or paper so it could be more cursive. I suppose that is something to consider when creating an "alphabet" for a fictional people.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts