Knowing when I'm done making grammar

If you're new to these arts, this is the place to ask "stupid" questions and get directions!
Post Reply
Cellular Automaton
rupestrian
rupestrian
Posts: 11
Joined: 08 Apr 2022 01:33

Knowing when I'm done making grammar

Post by Cellular Automaton »

Hey everyone! I'm working on my first conlang. I've made a phonology I'm pretty happy with, and I'm working on "reconstructing" the historical sound changes now, but I'm a little overwhelmed by the next steps: syntax and morphology. With phonology, I basically know what I need (consonants, vowels, phonotactics, allophony, stress rules), but with the grammar, I don't really have that kind of indication. So, how do I know what systems I need to set up and when I'm done with that? Do I just try to translate as much as possible and add in something new whenever I feel like I need it?
User avatar
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: Knowing when I'm done making grammar

Post by eldin raigmore »

fiziwig once posted a subset of “Graded Sentences for Analysis” and suggested that if you could gloss all of those sentences into your conlang, then your conlang was effectively “complete”. Or, at least, its grammar was. You might still need some vocabulary; but you probably had all your morphology and syntax worked out.
User avatar
sangi39
moderator
moderator
Posts: 3024
Joined: 12 Aug 2010 01:53
Location: North Yorkshire, UK

Re: Knowing when I'm done making grammar

Post by sangi39 »

My only problem with the Graded Sentences for Analysis approach is that the sentences, while good for looking at increasingly more complex sentences, lacks its own analysis (I don't know if the original book from the 1920s is any different, but I can't find a copy to read online to actually check), so you might run into the problem of using more and more "English-like" strategies as you go down the list if you can't figure out what's actually going on in those sentences, i.e. what's actually being encoded, as opposed to what words are being used in what order. I do think a lot of people get a lot of translation though, so it's definitely not a bad method, and it's always helpful to have sentences that other conlangers use to kind of compare and contrast what you're doing

Big one for me will always be to just read, even if it's just going through Wikipedia, WALS or the Universals Archive, bit by bit, having a bit of an exploration, get a few ideas and experiment with them, and then you can go to Google and try to find some more in-depth stuff, even language specific stuff, and really get a handle on what you want to do

Also, try not to rush, and give yourself to move around. If you get bogged down in one aspect, and it starts to grate on you a bit, there's always some other aspect you can move onto. Other than that, though, yeah, I guess "complete" is probably best thought of as being confident enough that when you come across a new sentence you can translate with relative ease (even if you still have to come up with some new vocab) as being happy with what you've got enough that you might tinker, but with no big overhauls
You can tell the same lie a thousand times,
But it never gets any more true,
So close your eyes once more and once more believe
That they all still believe in you.
Just one time.
Salmoneus
MVP
MVP
Posts: 3030
Joined: 19 Sep 2011 19:37

Re: Knowing when I'm done making grammar

Post by Salmoneus »

Hi.

The first thing to say is something you'll soon get tired of hearing if you ask conlangers many questions, because it's irritating and unhelpful, but it's also unfortunately true, and important to bear in mind: there is no right answer, no right way to do things. It all comes down to what you want to do. By some standards, you will never "be done" making grammar - a good grammar of English runs to hundreds and hundreds of pages and people are still debating and researching and filling in details. By other standards, you can be done in half an hour...

In terms of how to go about it, translation is really useful, yes, but is probably best not used on its own. The problem with having translation lead your grammar is that you can often end up not even noticing the grammatical questions a text raises, and so you don't write out those bits of the grammar, or you just copy directly from English (or another language you're familiar with) without being aware of the alternative options.

Example, spoilered for space:
Spoiler:
Here's a sentence I just thought of off the top of my head: Have you given it all to your brother yet, or do you still have some of it with you?
This seems pretty straightforward, but let's look at just a few of the grammatical issues that might not immediately be apparent to the naive translator:

- most of us will pretty quickly realise that "have you..." is not the only way a language could express a perfect tense, and that English's unusual patterns of verb inversion and auxiliary verbs need not be copied. But a language does need a way to form questions. And probably has some way to indicate a perfect, though in many languages this does not pattern as a tense (rather as an aspect, or a mood). And then there's a more difficult question: why does English use the perfect here? Or rather: why does English sometimes NOT use the perfect? ["did you give it all..." is a very similar alternative and it's hard to really specify the difference]
- let's note in passing the tense rules in "have you given..." (present auxiliary plus 'past' participle) vs "did you give..." (past auxiliary plus infinitive). Matching tenses between verbs is always a tricky bit of a language, and does your language even have an infinitive? Or a past participle?
- does it have auxiliaries? Can verbs ever be objects of verbs and if so do they have to be deverbalised in some way?
- what is "all" doing there? "All" normally looks like an adjective, but here it seems to be acting as an adverb - or is "it all" more like an apposition (like "Bob the Senator")?
- English has one word for 'brother'. Other languages distinguish younger from older brothers, or don't distingush brothers from cousins
- English has one word for "you", not distinguishing age or gender or social status
- English forms a possessive like "your brother" very easily - a noun preceded by what looks like an adjective. But some languages can't form "pronominal adjectives". Some languages don't let possession be marked by adjectives (or nouns in adjectivalised forms). Some languages do NOT distinguish possessive adjectives from genitive pronouns as English does (compare "Bob's brother" and "that is Bob's" with "your brother" and "that is yours"). Some languages mark pronominal and non-pronominal possession differently ("your brother" and "Bob's brother" wouldn't look alike). Some languages have more than one way to mark possession. Is a relative treated as the same sort of possession as a cactus or a spoon? Is a relative even an independent noun at all (in many languages it's a noun that can't exist without a possessive affix, or may even look a lot like a verb!... or there may be no noun at all and only a relative clause ("the one who brothers you")).
- let's not even go into the ways English uses "yet" and "still"!
- many languages don't have a simple word for "or" that combines independent clauses like this
- it seems beyond obvious that possession can be indicated with a possesive verb: "I have a pen". But in many languages, including the older forms of most European languages (Latin, Old English, modern Irish), this is not possible, or at least not the default (instead they use a structure along the lines of "a pen is with me")
- there won't necessarily be a single word that translates "some". If there is, what part of speech is it?
- "some of X" not only uses a prepositional phrase, or a sort more often used for physical position, but it uses the exact phrase that's also used for possession. Neither of these has to be true!
- does your language only have one word for "it"? Does it change depending on gender, number, animacy... or whether this is the first time you've mentioned it in your conversation?
- is there a simple word for "with" in all contexts? In English it has many meanings ("going with the flow", "eating with friends", "I have it with me", "a cat with stripes", "a problem with your answer", "you have to be careful with cats", etc - each of these could be distinguished, using a different preposition or even not a prepositional construction at all.

And so on. And that's just one random sentence I didn't plan!

------------

What alternatives are there to random translation? As mentioned, there are graded sentences - not only fiziwig's ones for conlangers, but you can probably find similar lists designed for assessing babies, languages learners or computers. These have the advantage of potentially putting their complexities front and centre, rather than having them lurk like unexploded bombs in the middle of a paragraph of inocuous prose for translation. They can encourage you to detect the 'problem' and think about it more than naturalistic text sometimes does. But they still have the problem of often leading the conlanger to automatically copy English. Sometimes even more so - because difficult sentences can go so deep into the complexities of a language, a conlanger is less likely to have thought about the issues they raise, and so more likely to reflexively copy what they know.

Another resources are questionnaires. WALS is a good questionnaire for natural languages, while CALS asks many of the same questions (and some others) of conlangs. There are also other questionaires used by linguists for describing and comparing languages in practice. These have the advantage of explicitly presenting you with 'problems' and showing you how other languages deal with them. But there are also downsides: they can encourage you to focus on certain areas at the expense of others, they can encourage conlangers to just throw in every 'strange' feature for the sake of it, and they can lead to disjointedness in conlanging, encouraging conlangers to see a language as a series of discrete and distinct questions, rather than as a cohesive system.

Finally, the best resource is just... other grammars. Of conlangs and of natlangs. Read about what other languages do. And not just modern European ones! There are also many linguistic articles on individual topics. Personally, although I think I lean toward a priori conlangs by instinct (conlangs not derived from real Earth languages), I've found that trying a posteriori conlanging has been hugely helpful in this regard. On of my major projects is a Germanic language - a cousin of English, derived from Proto-Germanic - and in particular the form of the language spoken in the late 1st millennium (comtemporaneous with Old English); this has encouraged me to read on Proto-Germanic, the various Germanic languages, and particularly Old English, and has really helped me in seeing the various ways my conlang can and should be unlike modern English.


--------


All that said, here are some ideas for things you should think about, beyond the most obvious topics...

- how do you handle valency (the number of arguments a verb takes)? Do you have zero-derivation (the same verb can have any number of arguments), or marked derivation (affixes to change the default number of arguments), or suppletion (different verbs for different valencies)? [in reality the answer is probably 'all three', but to differing extents]. When you add or subtract arguments, which arguments are being added or subtracted?

- how do you (or can you?) convey information about the extent to which a patient is affected by the verb, and does this vary depending on the type of verb, or the type of patient (or agent)? Options include dropping arguments, changing case assignments, adding marking to the verb, etc.

- how do you indicate which of two third-person referents is the agent and which is the patient, when you would otherwise want to use pronouns for both? ["The dog bit the man and then he ran away from him" - which one is running, the dog or the man?]

- how do you combine two clauses and make their aguments clear? [can you embed one clause as part of another, perhaps via nominalisation of the verb? can you chain one clause onto another while keeping it subordinated in some way? or are all clauses equal? How do you indicate that a referent in one clause is the same as the referent in an adjacent clause?] How do you indicate which clauses may be more important to the conversation and which provide background information?

- how do you deal with deixis - including (but not limited to) spatial deixis (here, there), ostensive deixis (this, that), contrastive deixis (if you like that try this!), proferative deixis (here, have this) and anaphoric (or cataphoric?) deixis (it, the aforementioned) and so on. With spatial deixis, for instance, how many degrees are distinguished and what defines them, and how do you determine what that deixis is relative to?

- how do you put focus onto an element of a clause? Does it differ depending on the type of focus? [eg contrastive, identificational, mirative]

- how do you indicate a topic of the conversation and what consequences does this have?

- how do you ask questions and make demands?

- how do you indicate other modal forces - how do you indicate that somebody ought to, could, presumably has, or failed to do something?

- how do you depict verbal aspects, and how does this interact with verbal classes and the inherent aspect of the verb?

- how does derivation interact with inherent word classes?


There are plenty of other questions you could ask, of course, but these seem like a few good ones to me.
User avatar
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: Knowing when I'm done making grammar

Post by eldin raigmore »

sangi39 wrote: 11 Apr 2022 19:10 My only problem with the Graded Sentences for Analysis approach is that the sentences, while good for looking at increasingly
….
have to come up with some new vocab) as being happy with what you've got enough that you might tinker, but with no big overhauls
It’s true IMHO that it might not be perfectly suitable or best for, say, polysynthetic-IV languages, or languages with “flat” structure, or with no clear distinction between coordinating and subordinating conjunction.
Nevertheless I think it might be good for checking whether you’ve left something out that just didn’t occur to you or just slipped your mind.
You may decide that certain examples from GSfA shouldn’t have very exact counterparts in your conlang, and that’s fine.

Most people who have written recommendations I’ve read, do recommend doing a lot of translating to grow your conlang.

And btw to be technical, no language is ever “complete” or “finished” until it’s dead. What one needs to ask oneself is whether it’s “complete enough” for the current purposes, whatever those are.

All IMHO! On re-reading it, this post doesn’t sound very humble. But try to imagine that I’m more humble than this makes me seem! I could be wrong, to a greater or lesser degree, about nearly any of this.
Last edited by eldin raigmore on 05 Aug 2023 03:12, edited 1 time in total.
User avatar
Sequor
sinic
sinic
Posts: 438
Joined: 30 Jun 2012 06:13

Re: Knowing when I'm done making grammar

Post by Sequor »

Cellular Automaton wrote: 09 Apr 2022 23:24but I'm a little overwhelmed by the next steps: syntax and morphology. With phonology, I basically know what I need (consonants, vowels, phonotactics, allophony, stress rules), but with the grammar, I don't really have that kind of indication.
I feel like I have that kind of indication in my head for grammar... More or less:

MORPHOLOGY (i.e. forms of words, purely speaking)
- inflectional: nouns, adjectives, verbs (conjugation), pronouns...
- derivational: to nouns, to adjectives, to verbs, to adverbs, numbers...

SYNTAX
- basic word order, structures of noun/verb phrases
- uses of inflectional categories (e.g. uses of the plural, uses of noun cases, uses of verbs' tense-aspect-moods...)
- possession (if not already covered under the uses of a genitive/dative case...)
- uses of determiners, demonstratives
- uses of quantifiers (all the, no X, numbers...), indefinite pronouns (everyone, some, nothing... maybe including related words like somewhere and always/never)
- noun clauses (he says that we should do...)
- relative clauses (that, which, the one that..., who did..., what I did)
- list of prepositions and their uses (before, during, in, for...)
- adverbial phrases and their conjunctions (before, while, when, although, so that...)
- (same-level) conjunctions (and, but...)
- reflexives and reciprocals (do something to oneself, do something to each other...)
- conditions
- comparisons (bigger than, as big as...)
- topicalization and focus
- auxiliaries typically including modals (might do, can do...), and linked verbs in general (keep doing, try to do, hope to do...)
- (transforming clauses into) questions, exclamations
- negation (i.e. transforming clauses into negated ones)
- passives (including transforming clauses into passive ones)
- valency (how the conlang deals with verbs that can vary in what arguments they take)
- contextual reduction (omitting words to avoid repetition in certain constructions, notably in comparisons, answers to questions...)

Obviously, this list works better for languages that aren't morphology-heavy let alone polysynthetic. I like to separate wordforms (morphology) from usage, but you don't need to do that really. And regardless it must be adapted to fit the needs of your conlang. Lots of European natlang grammars like to have a subchapter or chapter for each TAM (verb tense-aspect-mood), for example. There's at least one Chinese grammar (Yip and Rimmington's, 2nd edition) that dedicates a whole chapter to compound words, classifying the many ways in which they can be formed. Some grammars separate modifier relative clauses (the book that I brought) from "headless" relative clauses (what I brought). And so on.

Also, in practical terms, it is necessary to start small and accumulate detail as you go. There's no point in feeling overwhelmed by all the work that can be done. There's no end to the grammar description that can be done. Also, lots of conlangers perfectly get by with grammars that don't cover some of the above in any level of detail, whether they carry their decisions in their head, or simply calque the constructions of a language they're familiar with...
Last edited by Sequor on 09 Oct 2022 16:23, edited 2 times in total.
hīc sunt linguificēs. hēr bēoþ tungemakeras.
Khemehekis
mongolian
mongolian
Posts: 3883
Joined: 14 Aug 2010 09:36
Location: California über alles

Re: Knowing when I'm done making grammar

Post by Khemehekis »

If you'll allow me to toot my own horn, Cellular Automaton, here's the grammar I wrote for my best-developed conlang, Kankonian.

https://khemehekis.angelfire.com/basic.htm

You can start with the things Sequor lists, then move on to the kind of things Salmoneus mentions, and once you've gotten all of those taken care of, you can check out the Kankonian grammar and see if there's anything there that you still don't have covered (how do you express SAT-style analogies in your conlang, for instance?)

Also recommended is the Snowball Game, on this board, for all sort of grammatical acrobatics.
♂♥♂♀

Squirrels chase koi . . . chase squirrels

My Kankonian-English dictionary: 86,336 words and counting

31,416: The number of the conlanging beast!
User avatar
MissTerry
hieroglyphic
hieroglyphic
Posts: 58
Joined: 14 Apr 2022 02:23

Re: Knowing when I'm done making grammar

Post by MissTerry »

You are never truly done with the grammar. A living language - like anything living - is always in a gradual/incremental state of change and growth. For example, the English language of today and the English language of the 1500s are very different from each other. The English we see in the traditional King James Bible has words, affixes, and changes/morphology that we don't use today [Thou, Doth, Wherefore Art Thou Romeo?]. There was a time when I was in grade school when the teacher would reprimand us for pluralizing the Latinate word "Cactus" as "Cactuses" and not as "Cacti." But today, adding the very Englishie suffix +ES to a Latin loanword "cactus" is acceptable vernacular grammar.

I keep my conlang "opened" to incremental grammatical change as the need or requirement arises. For example, my nouns have a simple and straightforward grammar for them, where all nouns end in +A. So "Druya" = Tree [Druya derives from Sanskrit]. The dual plural suffix is +I. So "Druyi" = Two Trees or a pair of trees. The plural suffix is +AI. So "Druyai" = Trees. And then +U = "The." And so "Druyu" = The Tree. That's all fine, theoretically. But in practice, I ran into a non-aesthetically pleasing issue: how do I say "The Trees?" Druyaiu sounds stupid. And Druyiu meaning "The Pair of Trees" sounds just as dumb. And so, to fix that aesthetic issue, I further developed my grammar to include 2 new definitive articles: Di & Tu. Di means "The" but indicates "Two of something." Tu also means "The" but "many somethings." And so Druyu = The Tree; Di Druya = The Pair of Trees; & Tu Druya = The Trees.

Rules of grammar is like sandpaper, which helps smooth and streamline your language. A living language is never done refining itself [across the span of time]. And so, a conlang beyond fancy theories and glossing fetishes, should never stop evolving and streamlining itself. Old words, clunky words atrophy from disuse. New words are gradually added or borrowed or coined. With new words comes new incremental changes in grammatical rules. Relatively recently, English borrowed the German word "Weltanschauung," and so with the acquisition of that new word, comes grammar rules associated with that word: currently, the proper/educated way to pluralize that loanword is to follow the German practice: Weltanschauungen," not "weltanschauungs."

A living language, in time, develops to be more precise in explaining power, and precise with lesser words as a previous evolutionary iteration of that language. For example, instead of saying 2000 Pounds, we say a "ton." Instead of saying: The self-driving carriage weighs 12,000 pounds; we can today with the same English language say: the automobile weighs 6 tons. And with the innovation of the neologism of "automobile" comes needed minor grammar rules/practice: may we shorten that neologism? If we may shorten that neologism, which part may we use? "Mobile?" or "Auto?" If we call a vehicle a "mobile," then what would we call a mobile cellular phone? If we call it an "auto," then how should we pluralize it? According to Greek tradition as "Autoi," or do we add the traditional English +S making the word "Autos?"

A living conlang should exist beyond theory. In practice, the point of a language is to communicate and explain things. In that endeavor, precision is the gradual direction of a language's trajectory of evolution. Today a language like English has become more powerful in its ability to precisely explain the universe: thanks mostly to science, and scientific neologisms. With such new words comes tweaks of grammar into order to streamline and smoothen the English. Should 'black hole' be one word or two? Should a hyphen be used like this: black-hole. If the word "Molecule" ends with a +Cule, and the old word "Corpscule" ends also with a +Cule, then can we grammatically use that ending as a suffix and append it to other English words, like a Quarkcule, or a Bose-Einsteinian Condensatecule? We do with do that with the +Gate in "Watergate," where we make the "Gate" into a suffix which tries to mean something scandalous.

A living conlang, one that exists beyond theory and paper, should continue to incrementally evolve/develop towards more powerful capacity to communicate and explain, and to have greater power of precision in communication and explaining stuff. With that incremental development, comes atrophy of old clunky words, inventions/borrowing/innovations of new words or neologisms. With that comes the need for tweaks in grammar to streamline and smoothen the new iteration. And so the process is continuous and doesn't ever really stop. There are of course languages that do not evolve or change in grammar: Latin [used in science to name plants and animals], Sanskrit [used in mantras], and Pali [the sacerdotal language of Theravada Buddhism] for instance. But those languages are Dead Languages, they are textual/written languages [a language which exists only on paper, in text form].
Khemehekis
mongolian
mongolian
Posts: 3883
Joined: 14 Aug 2010 09:36
Location: California über alles

Re: Knowing when I'm done making grammar

Post by Khemehekis »

Also: You may be interested in the Fink-Peterson test sentences . . . same basic idea as the Snowball Game.
♂♥♂♀

Squirrels chase koi . . . chase squirrels

My Kankonian-English dictionary: 86,336 words and counting

31,416: The number of the conlanging beast!
User avatar
lsd
greek
greek
Posts: 740
Joined: 11 Mar 2011 21:11
Contact:

Re: Knowing when I'm done making grammar

Post by lsd »

I don't think it's necessary to write a grammar,
I've been maintaining the same language for many years, and I haven't written a line....

but it is necessary to have a mental grammar of the language, the one that is acquired through use.....
for a language without a speaker, it is necessary to produce...

a diary if you like that,
or translate which is the king of conlanging,
so why not the standard sentences,
I prefer to describe my environment, and to translate what passes within my reach...
Post Reply