How to begin an a priori conlang
-
- Sanci
- Posts: 34
- Joined: Sun Nov 24, 2013 4:50 am
How to begin an a priori conlang
When making a priori conlangs, usually you begin with a proto-language and then derive the daughter. But how do you make a mother without first making another mother? I can't think of ways to properly construct a conlang without prior precedent, which for me (who usually try to make a priori ones) is a very bad thing.
-
- Lebom
- Posts: 204
- Joined: Thu Feb 25, 2010 7:26 pm
Re: How to begin an a priori conlang
Turtles all the way down.
Realistically though look at it like PIE. We know that's not the earliest form, but we can't realistically recreate anything much earlier, so at some point you just have to say, I'm going to make this stuff up for this ancestor language and then use the rules of derivation and sound change for all its descendants.
Realistically though look at it like PIE. We know that's not the earliest form, but we can't realistically recreate anything much earlier, so at some point you just have to say, I'm going to make this stuff up for this ancestor language and then use the rules of derivation and sound change for all its descendants.
Re: How to begin an a priori conlang
You could create a hypothetical proto-language thousands of years before the oldest stage you want to have a full reconstruction for, and pretend that that proto-language was 100% isolating and had a perfectly symmetrical phonology. The explanation would be that any grammatical complexity that it did have was lost in all the daughter languages. That is what we do with real-world examples in some cases when there is only one attested daughter language. I mean for all we know, Proto-Indo-European had ejective uvulars all over the place but went through a sound change shortly after splitting up which changed them all into a boring /k/ sound. (This is a serious proposal, which because it's impossible to disprove, can never be ruled out, but in my opinion does not seem very likely.)
I think a good stopping point would be to go to your current proto-language, i.e. the one language everything evolves from, and then go back two more "generations" (probably about 2000 years) making a hypothetical pre-proto-language that had only that one descendant, so that you wont have to worry about adding more features to the pre-proto-language to account for the late proto-language's sister languages. Analogous to how people reconstructing pre-proto-PIE dont worry about what features might have been lost in Pre-Proto-PIE as it evolved into standard PIE.
I take my own advice. I'm currently looking at a table of sound changes covering roughly 9200 years of my conworld, and I've got ideas in my head that I haven't written down yet that go back even a thousand or so before that, but for the grammar of those languages I have no real conception other than just mirroring back the grammars of the languages they evolve into. So I can tell you how to say "girl" and "seashell" and "sit down" in pre-proto-Silatibarra, but not how to say "the girl sat down on the seashell."
I think a good stopping point would be to go to your current proto-language, i.e. the one language everything evolves from, and then go back two more "generations" (probably about 2000 years) making a hypothetical pre-proto-language that had only that one descendant, so that you wont have to worry about adding more features to the pre-proto-language to account for the late proto-language's sister languages. Analogous to how people reconstructing pre-proto-PIE dont worry about what features might have been lost in Pre-Proto-PIE as it evolved into standard PIE.
I take my own advice. I'm currently looking at a table of sound changes covering roughly 9200 years of my conworld, and I've got ideas in my head that I haven't written down yet that go back even a thousand or so before that, but for the grammar of those languages I have no real conception other than just mirroring back the grammars of the languages they evolve into. So I can tell you how to say "girl" and "seashell" and "sit down" in pre-proto-Silatibarra, but not how to say "the girl sat down on the seashell."
And now Sunàqʷa the Sea Lamprey with our weather report:
Re: How to begin an a priori conlang
To be fair, the hypothesis that the Proto-Indo-European dorsal stops places of articulation are velar, uvular, and labialized uvular rather than palatal/palatalized velar, plain velar, and labialized velar wasn't conjured out of thin air. If the "palatal" stops were palatalized, they would have had to uniformly de-palatalize in the centum languages, which is a highly unusual and extremely uncommon sound change, and we would expect exceptions to occur in at least some of the daughter languages. Instead, the evidence we have seems to suggest that satemization, which turned the "palatal" stops into some manner of affricate or fricative, was a secondary change that occurred late in the history of PIE, with daughter families as widely dispersed as Anatolian, Germanic, Hellenic, and Tocharian unaffected.Publipis wrote:I mean for all we know, Proto-Indo-European had ejective uvulars all over the place but went through a sound change shortly after splitting up which changed them all into a boring /k/ sound. (This is a serious proposal, which because it's impossible to disprove, can never be ruled out, but in my opinion does not seem very likely.)
The most parsimonious explanation seems to be that PIE had some kind of post-velar series which merged with the velar series in most languages (not a terribly unlikely change, especially considering that there were a lot more velars than post-velars), but that the Satem langauges instead underwent a chain shift that pushed the plain velar series to a more palatal place of articulation.
Of course, this scenario isn't without it's problems; for one thing, it requires the geographically unrelated centum languages to uniformly lose their pre-velar series. Although this is more likely than them all undergoing depalatalization, some people think it's a compelling reason to think that PIE had only two stop series to begin with. However, if there were only two series, it raises the question of why most plain velars were palatalized in the satem languages while a minority were not. There doesn't seem to be any uniform conditioning factor that can be pointed to as an explanation. Furthermore, Luwian, part of the Anatolian family (probably the earliest to have branched off from the "core" Indo-European languages) seems to preserve three different reflexes for the dorsal series: *ḱ becomes "z" (probably an affricate like [ts]), *kʷ becomes "ku" (probably [kʷ]), and *k becomes [k].
Re: How to begin an a priori conlang
Do... do you mean a priori or do you mean a posteriori
Because most people when they're making an a priori language just, you know, make it - in fact making an a priori language developed from a proto-language (albeit an a priori one) kind of means you're already into a posteriori
Because most people when they're making an a priori language just, you know, make it - in fact making an a priori language developed from a proto-language (albeit an a priori one) kind of means you're already into a posteriori
كان يا ما كان / يا صمت العشية / قمري هاجر في الصبح بعيدا / في العيون العسلية
tà yi póbo tsùtsùr ciivà dè!
short texts in Cuhbi
Risha Cuhbi grammar
tà yi póbo tsùtsùr ciivà dè!
short texts in Cuhbi
Risha Cuhbi grammar
- prettydragoon
- Sanci
- Posts: 67
- Joined: Sat Jan 07, 2012 4:31 pm
- Location: Haru
Re: How to begin an a priori conlang
Essentially, if you are making an a priori conlang, that implies you have to pull most of it e posteriori.
Re: How to begin an a priori conlang
I'm not at all sure making a protolang first can be said to be the most usual methodology for apriori by any stretch of the imagination: *actually* it is not, i claim, even mathematically possible for it to be the case that 'most apriori langs are done by first doing a protolang'.PVER•PVERUM•AMAT wrote:When making a priori conlangs, usually you begin with a proto-language and then derive the daughter. But how do you make a mother without first making another mother? I can't think of ways to properly construct a conlang without prior precedent, which for me (who usually try to make a priori ones) is a very bad thing.
Instead, you just, you know, come up with stuff.
Re: How to begin an a priori conlang
edit: sorry my lack of schoolboy latin stopped me from getting this excellent jokeprettydragoon wrote:Essentially, if you are making an a priori conlang, that implies you have to pull most of it e posteriori.
كان يا ما كان / يا صمت العشية / قمري هاجر في الصبح بعيدا / في العيون العسلية
tà yi póbo tsùtsùr ciivà dè!
short texts in Cuhbi
Risha Cuhbi grammar
tà yi póbo tsùtsùr ciivà dè!
short texts in Cuhbi
Risha Cuhbi grammar
-
- Niš
- Posts: 1
- Joined: Tue Sep 30, 2014 1:57 am
Re: How to begin an a priori conlang
Okay, I'm a total noob at this and I'm not actually quite sure how to begin in everything. So with that, where should I go in this forum if say I wanted to know the basics of conlang? Thanks.PVER•PVERUM•AMAT wrote:When making a priori conlangs, usually you begin with a proto-language and then derive the daughter. But how do you make a mother without first making another mother? I have been looking at some sites like this but I could not find the answer. It is important for me to find the answer. I can't think of ways to properly construct a conlang without prior precedent, which for me (who usually try to make a priori ones) is a very bad thing.
Last edited by ruebentucci55 on Mon Mar 20, 2017 9:22 pm, edited 1 time in total.
Re: How to begin an a priori conlang
there's a book, called the language construction kit... there's also a webpage that goes by the same name and has a lot of what you need to get started. There exists also other instruction sets, such as one made by pablo flores iirc, but that one's in spanish. I'm sure there's other ones going around.
The following is my way, other guys and gals each do it differently.
* decide on the general shape and typological coordinates of your lang: do you want it to be like a highly polysynthetic lang with free-flowing word order and a lot of complicated morphology? do you want it to be kind of like english, or chinese, where individual words inflect little or not at all and where word order and particles convey most of the meaning? do you want it to be romance-sounding and similar to, say, italian? or do you want something full of weird ass ejectives? do you want adjectives as a separate noun class? SVO syntax? if you don't know what these mean, no worries. just make a few notes along the lines of what you want your lang to be and go forward in peace with the grace of the lord.
* decide on general phonetic traits: what do you want the thing to sound like. jumbled and strange like tamazigh? choppy japanese? machinegun-like a la spanish? rhytmic like brazilian portuguese? play around with your mouth and the sounds it can make. decide on whether you want tones or something like that.
* pick up a vowel system <that is to say, choose whether your lang will have aeiou or aie or the twenty-something vowels of english, or maybe the six vowels or chinese.
* pick up a consonant system: maybe you want something complex, ptk?shmnlr? maybe you want a more complex, fricative-heavy one, like rnstlčgxdmθbfpßðkȝwɣ. maybe more complicated? like http://en.wikipedia.org/wiki/Kabardian_ ... Consonants or maybe you want something more wack than crack, in which case you can pick up inspiration here http://en.wikipedia.org/wiki/Abkhaz_language#Phonology. venture into more complex consonant systems at your own peril, for that way lies madness.
* figure out what kind of phonotactics you want: you want strict ones, like CONSONANT-VOWEL-CONSONANT-VOWEL all the way, or you want strange and jumbled wordforms like sprite or mpontna ?
* make a shitload of words, either by hand or using wordgens. there exist many, but awkwords works for most purposes: hell, i rarely use anything but
* apply a few sound changes to the bunch of words you have to play around with the frequencies and the sound: maybe you want to turn all instances of uu into wu as long as they're before a consonant, and turn them into plain w if they're before a vowel, maybe you want to restrict dipthongs to just wa and ja and turn all the other dipthongs into monopthongs, whatever floats yer boat.
once you have a sound and a shitload of words, you need to figure out grammar.
The order in which this is done will depend, i claim, in the kind of language you want to do: if you went for an uninflected lang where grammar happens in word order and particles, you can skip morphology [we like to call those langs isolating]. if you want free-flowing word order, you're definitely gonna need some morphology. Morphology is just fancyspeak for how words change to reflect stuff. for example, penis changes to penises to reflect that the speaker means more than one penis: this can be thought of as there's this word "penis" which takes a suffix "-es" to reflect plural. so the plural of penis is penises <yeah, yeah, this is not a good analysis of english plurals>. There's a shitload of kinds of information words can change to reflect: for example, in castillian doctor can change to doctora to communicate that the person you're speaking is indeed a doctor with tits. or a female doctor without them. In a conlang of mine, nouns reflect the time they're at, so if i were to speak of my dead granny i'd have to use something like the past perfect, since my dead grandmother is no longer, she already was. Maybe your verbs reflect whether the verb was done on purpose. there's also many ways in which words can inflect: for example, some words take suffixes, like penis|penises. other words take prefixes, other circumfixes, other use non-concatenative morphology, blablabla. Here's where a lot of conlanging is done: do you have case? fine, what's the morphosyntactic alignment of your lang?
Also you need to figure out the general syntax: is it subject-verb-object? maybe object-subject-verb? maybe there's no verbs and you express actions and changes of state in new and wonderful ways? what is the structure of noun phrases? what about prepositions, do you have them?
Then you need to figure out a bunch of core lexicon. languages will generally have a number of special, low-content high-prequency words; you know, and, or, yes, without, my, yours, much, under, below, over, no: what are they? <if your lang doesn't have any articles, classifiers, determiners, comparators, or any other such things more power to you, you'll have to express the concept of "and" or "below" in some way or other. maybe suffixes, like with a specific case of "and" and latin>.
At one point, you need to start using your lang to say stuff: just take whatever text and translate it into your conlang: if you're like me, one or two sentences in you're gonna say "jesus, how does [whatever stuff] works?man, i need to go back to the drawing board and figure out how this language does [greetings, relative clauses, comparatives, whatever]
I express these things linearly but i'm lying: you're gonna be going back and forth between issues: oh, did you just decide the language has noun class? well, that's gonna affect the morphology of your pronouns [or not, i don't know]. and so on and so on
If you're anything like me, you'll eventually lose interest in conlanging, come back to the board for shits and giggles, until the itch strikes again, in which case you'll start all over again. You're one of us now, you'll never stop conlanging. never mwahahahaha.
The following is my way, other guys and gals each do it differently.
* decide on the general shape and typological coordinates of your lang: do you want it to be like a highly polysynthetic lang with free-flowing word order and a lot of complicated morphology? do you want it to be kind of like english, or chinese, where individual words inflect little or not at all and where word order and particles convey most of the meaning? do you want it to be romance-sounding and similar to, say, italian? or do you want something full of weird ass ejectives? do you want adjectives as a separate noun class? SVO syntax? if you don't know what these mean, no worries. just make a few notes along the lines of what you want your lang to be and go forward in peace with the grace of the lord.
* decide on general phonetic traits: what do you want the thing to sound like. jumbled and strange like tamazigh? choppy japanese? machinegun-like a la spanish? rhytmic like brazilian portuguese? play around with your mouth and the sounds it can make. decide on whether you want tones or something like that.
* pick up a vowel system <that is to say, choose whether your lang will have aeiou or aie or the twenty-something vowels of english, or maybe the six vowels or chinese.
* pick up a consonant system: maybe you want something complex, ptk?shmnlr? maybe you want a more complex, fricative-heavy one, like rnstlčgxdmθbfpßðkȝwɣ. maybe more complicated? like http://en.wikipedia.org/wiki/Kabardian_ ... Consonants or maybe you want something more wack than crack, in which case you can pick up inspiration here http://en.wikipedia.org/wiki/Abkhaz_language#Phonology. venture into more complex consonant systems at your own peril, for that way lies madness.
* figure out what kind of phonotactics you want: you want strict ones, like CONSONANT-VOWEL-CONSONANT-VOWEL all the way, or you want strange and jumbled wordforms like sprite or mpontna ?
* make a shitload of words, either by hand or using wordgens. there exist many, but awkwords works for most purposes: hell, i rarely use anything but
* apply a few sound changes to the bunch of words you have to play around with the frequencies and the sound: maybe you want to turn all instances of uu into wu as long as they're before a consonant, and turn them into plain w if they're before a vowel, maybe you want to restrict dipthongs to just wa and ja and turn all the other dipthongs into monopthongs, whatever floats yer boat.
once you have a sound and a shitload of words, you need to figure out grammar.
The order in which this is done will depend, i claim, in the kind of language you want to do: if you went for an uninflected lang where grammar happens in word order and particles, you can skip morphology [we like to call those langs isolating]. if you want free-flowing word order, you're definitely gonna need some morphology. Morphology is just fancyspeak for how words change to reflect stuff. for example, penis changes to penises to reflect that the speaker means more than one penis: this can be thought of as there's this word "penis" which takes a suffix "-es" to reflect plural. so the plural of penis is penises <yeah, yeah, this is not a good analysis of english plurals>. There's a shitload of kinds of information words can change to reflect: for example, in castillian doctor can change to doctora to communicate that the person you're speaking is indeed a doctor with tits. or a female doctor without them. In a conlang of mine, nouns reflect the time they're at, so if i were to speak of my dead granny i'd have to use something like the past perfect, since my dead grandmother is no longer, she already was. Maybe your verbs reflect whether the verb was done on purpose. there's also many ways in which words can inflect: for example, some words take suffixes, like penis|penises. other words take prefixes, other circumfixes, other use non-concatenative morphology, blablabla. Here's where a lot of conlanging is done: do you have case? fine, what's the morphosyntactic alignment of your lang?
Also you need to figure out the general syntax: is it subject-verb-object? maybe object-subject-verb? maybe there's no verbs and you express actions and changes of state in new and wonderful ways? what is the structure of noun phrases? what about prepositions, do you have them?
Then you need to figure out a bunch of core lexicon. languages will generally have a number of special, low-content high-prequency words; you know, and, or, yes, without, my, yours, much, under, below, over, no: what are they? <if your lang doesn't have any articles, classifiers, determiners, comparators, or any other such things more power to you, you'll have to express the concept of "and" or "below" in some way or other. maybe suffixes, like with a specific case of "and" and latin>.
At one point, you need to start using your lang to say stuff: just take whatever text and translate it into your conlang: if you're like me, one or two sentences in you're gonna say "jesus, how does [whatever stuff] works?man, i need to go back to the drawing board and figure out how this language does [greetings, relative clauses, comparatives, whatever]
I express these things linearly but i'm lying: you're gonna be going back and forth between issues: oh, did you just decide the language has noun class? well, that's gonna affect the morphology of your pronouns [or not, i don't know]. and so on and so on
If you're anything like me, you'll eventually lose interest in conlanging, come back to the board for shits and giggles, until the itch strikes again, in which case you'll start all over again. You're one of us now, you'll never stop conlanging. never mwahahahaha.
Re: How to begin an a priori conlang
Maybe the prominent sticky entitled For new members in None of the above? (Actually, being honest, it's not the most intuitive place for such a sticky- is there any way that phpbb would allow linking from the "front page" of the forum, as it were?)ruebentucci55 wrote:Okay, I'm a total noob at this and I'm not actually quite sure how to begin in everything. So with that, where should I go in this forum if say I wanted to know the basics of conlang? Thanks.PVER•PVERUM•AMAT wrote:When making a priori conlangs, usually you begin with a proto-language and then derive the daughter. But how do you make a mother without first making another mother? I can't think of ways to properly construct a conlang without prior precedent, which for me (who usually try to make a priori ones) is a very bad thing.
Salmoneus wrote:(NB Dewrad is behaving like an adult - a petty, sarcastic and uncharitable adult, admittedly, but none the less note the infinitely higher quality of flame)
- Curlyjimsam
- Lebom
- Posts: 205
- Joined: Wed Dec 29, 2004 11:57 am
- Location: Elsewhere
- Contact:
Re: How to begin an a priori conlang
A question that I think deserves asking is "What are the main advantages of creating an ancestor language?". I can think of a few main ones:
(1) Being able to produce historical texts in a language before the "modern" period. If you want to do this, you need to be able to trace your language back at least to the point where it was first written down, but not necessarily any further.
(2) Being able to produce related languages. In this case, you need to be able to go back to at least the point where the proto-language "split" into two or more daughters.
(3) Creating irregularities that nevertheless exist for clearly definable reasons. (For example: "verbs x,y,z are irregular in a particular way W because they were affected by a particular sound change S".) An important thing to note about irregular forms, though, is that they tend to be unstable, particularly over extremely long periods. For example, there are very few irregularities in English that can be traced back to PIE, I believe. This is helpful for the conlanger because it means that if you go far enough back there's really very little point in detailing any irregularities in your proto-language as they won't survive to any descendant(s) that exist in the modern-day or were written down in the past anyway. This is turn means you reach a point where there's little reason to detail a language's history for this purpose.
(4) Creating particular sound patterns. For example, there is a distinctive pattern in my language Viksen where non-initial stops tend to be voiced, for reasons relating to historical sound change. This and other such patterns lends it quite a distinctive flavour that probably wouldn't be so strong if I'd just made up the words myself. But again, once you go back far enough any such patterns which might have existed in an ancestor are liable to have been eroded by the "present", so you perhaps don't have to worry about them too much. Additionally, this sort of thing is quite easy to fake (i.e. you can create patterns that look like they might result from particular changes without ever bothering to think too much about those changes in detail), especially if the trends aren't absolute.
So I would say you will always reach a point where going back further in time is basically futile.
(1) Being able to produce historical texts in a language before the "modern" period. If you want to do this, you need to be able to trace your language back at least to the point where it was first written down, but not necessarily any further.
(2) Being able to produce related languages. In this case, you need to be able to go back to at least the point where the proto-language "split" into two or more daughters.
(3) Creating irregularities that nevertheless exist for clearly definable reasons. (For example: "verbs x,y,z are irregular in a particular way W because they were affected by a particular sound change S".) An important thing to note about irregular forms, though, is that they tend to be unstable, particularly over extremely long periods. For example, there are very few irregularities in English that can be traced back to PIE, I believe. This is helpful for the conlanger because it means that if you go far enough back there's really very little point in detailing any irregularities in your proto-language as they won't survive to any descendant(s) that exist in the modern-day or were written down in the past anyway. This is turn means you reach a point where there's little reason to detail a language's history for this purpose.
(4) Creating particular sound patterns. For example, there is a distinctive pattern in my language Viksen where non-initial stops tend to be voiced, for reasons relating to historical sound change. This and other such patterns lends it quite a distinctive flavour that probably wouldn't be so strong if I'd just made up the words myself. But again, once you go back far enough any such patterns which might have existed in an ancestor are liable to have been eroded by the "present", so you perhaps don't have to worry about them too much. Additionally, this sort of thing is quite easy to fake (i.e. you can create patterns that look like they might result from particular changes without ever bothering to think too much about those changes in detail), especially if the trends aren't absolute.
So I would say you will always reach a point where going back further in time is basically futile.
Re: How to begin an a priori conlang
Yes, if you want to make the language look like it was derived from something else, it's nice to have something to derive it from, essentially making it an a posteriori language, but based on another conlang. But at some point you're going to have to just make things up. I guess if you're not too picky about the parent (or grandparent) language, you can make some use of generator programs. It might even not matter so much if you happen to copy natlang semantic categories, if it's for a protolang you're not really going to use.