-1

For example, the different types of language:

  1. Pidgin
  2. Creole
  3. Standard natural language
  4. Conlang

It seems we could create a stepping-stone map going from pidgin -> "stereotypical natlang" somehow. Somehow there is a process of creating the language, which is very murky / blurry to my mind currently. The easy part is phonetics/phonology, and lexicon. The super hard part is the sentence structures, in my experience so far.

Side question, are any "natlangs" (what you'd think of as a natural language traditionally) actually conlangs? Like Sanskrit? Was it made up by an individual or small group? What about Korean or Tibetan, I know those writing systems were supposedly created by one person (same with Armenian if I remember correctly). Or is that just writing systems, not the language constructed?

I have been working on a conlang for at least a year, having mostly developed a lexicon and basic rules for simple noun and verb phrases (seen in that link's website). I have played a lot with sentence structures, absorbing as much as I can from the Chinese Grammar Wiki "sentence patterns" / templates, and my knowledge of English. I don't know Chinese well, it will take me a few years to learn it enough to move it into working/long-term memory.

I am still stuck on mostly mirroring English words and sentence patterns, because that is most familiar. I have made this language maximally isolating/analytic, sort of like Vietnamese/Chinese, where everything is its own word (no conjugations or inflections or anything). Even past/future tenses are words, as is pluralizing, etc..

My question is, how do conlangers iterate on their sentence structure? To me, it seems like you must first start speaking a prototype version of your language, then expanding/refining its sentence structures as you feel what works and what doesn't. This would inherently require memorizing your lexicon of at least a few hundred words, and starting to try and "think" in terms of your language.

Can you paint a picture of how I might go about creating sentence structures and formalizing the grammar, without knowing things in advance, so I can make a bit more progress?

How do natural languages handle this? Or, since no one knows, how might they have evolved to their current state of complexity, in what sort of phases or steps? A conlang might seem to be a microcosm of how a natural language develops, sped up in time and developed by an individual or small group instead of a relatively large community.

How do you figure out sentence structures?

  1. Create some arbitrary sentence structures, perhaps based on some other language (like English). Say 10 types of sentences, or 10 features of sentences.
  2. Try and speak and understand what you're saying (as a solo conlanger) using those structures.
  3. When it doesn't make sense, refine? That is where it become super fuzzy to me... How do you refine your sentence structures?

I get caught up in the process of testing out sentences using my rules, and not wanting to memorize my language rules without them being finalized. So somehow I need to try and speak, but evolve the rules as I go. Whereas in a natlang in today's world, the rules are pretty much set in stone in advance.

How do you create rules which are flexible and evolving, and not set in stone at first, and still make process in producing an understandable language?

P.S. if you have any books or papers which are relevant here, please add a comment, I would love to take a deeper look.

2
  • 2
    I say this on most of your questions, but the answer is really, really, to learn at least one language that is not English. Then you'll get used to some of the different ways of structuring sentences and how they can vary, instead of just copying what English does.
    – Draconis
    Aug 1, 2023 at 17:28
  • This line of research (following the Bickerton trail) seems interesting, the transition from "protolanguages" to "true" languages.
    – Lance
    Aug 2, 2023 at 4:14

1 Answer 1

2

You have a legitimate question as to how to build sentences, right?

Your question, albeit a bit verbose, is a legit question which I interpret as you wanting to know how to build sentences that don't copy English. If I have misinterpreted your question, I hope my response is nonetheless useful to you.

So! Let me give a few tips which really helped me. I'll be citing the accredited World Atlas of Language Structures (WALS) throughout my response.

Firstly, I like to build sentence structure from its components rather than choosing SOV, SVO, VSO... that way I can get a better feel and intuition of my conlang and get a sense of how naturalistic it is. Fortunately, we don't need to know conlang vocabulary to do this.

Order of Object and Verb

A lot can be inferred about a language based on its order of object and verb which is why I favor deciding this first when I make conlangs (more on this later). WALS Feature 83A lists some useful statistics which are

  • 712 of 1518 (46.9%) surveyed languages favor OV,
  • 705 of 1518 (46.4%) surveyed languages favor VO, and
  • 105 of 1518 (6.65%) surveyed languages favor no particular order.

As seen here, the world's languages are pretty evenly spit between OV and VO, so just choose one that you like. Consider the implications of each and why your fictional speakers would orient their syntax in such a manner. Additionally consider the effect of animacy. The sentence "man ate berries" is unambiguous in any order because berries aren't known for eating. But the sentence "bird ate snake" is ambiguous without case marking or other methods since birds can eat snakes and snakes can eat birds.

Subject Location

By far the most common order for subject and verb is, well, subject then verb. 1192 of 1496 (79.7%) surveyed languages have SV order (see: WALS Feature 82A). When compared against the order of object and verb, we see that

  • 457 of 692 (66.0%) of surveyed VO languages favor SV,
  • 180 of 692 (26.0%) of surveyed VO languages favor VS,
  • 676 of 695 (97.3%) of surveyed OV languages favor SV, and
  • 7 of 695 (1.01%) of surveyed OV languages favor VS.

Some languages have fluidity in their syntax, and these are accounted for in the unlisted percentages.

Adposition Location / Order of Adposition and Noun Phrase

Again, citing WALS to compare feature 83A against feature 95A,

  • 456 of 498 (91.5%) of surveyed VO languages favor prepositions
  • 472 or 486 (97.1%) of surveyed OV languages favor postpositions

Like before, you as the conlanger get to choose which you like. Sound it out in your head (or aloud if you so desire) to decide whether you want sentences like

  • "Shepherds in Spain watch sheep", or
  • "Shepherds Spain in watch sheep", or
  • "In Spain shepherds watch sheep", or
  • "Spain in shepherds watch sheep", or
  • something of your own design.

Try not to analyse these with an English perspective. Try to put yourself into the paradigm of your conlang's speakers. Perhaps think poetically? Find what works for you.

Adjective Location / Order of Adjective and Noun

One way you can differentiate your conlang from English is the position of adjectives and nouns. English places adjectives before nouns, but 64.3% of languages favor the opposite! (see: WALS Feature 87A) Comparing against order of Object and Verb, we see that

  • 456 of 612 (74.5%) of surveyed VO languages favor Noun-Adjective,
  • 114 of 612 (18.6%) of surveyed VO languages favor Adjective-Noun,
  • 332 of 597 (55.6%) of surveyed OV languages favor Noun-Adjective, and
  • 216 of 597 (36.2%) of surveyed OV languages favor Adjective-Noun.

Knowing the order of adjective and noun is useful for knowing how words will tend to compound and form genitives. Since adjectives precede nouns in English, words compound in that order (e.g. saucepan).

Possessor and Possessee / Order of Genitive and Noun

Conlang YouTuber Biblardion in his How to Make a Language series suggests that the order of Adjective and Noun is strongly related to the order of Genitive and Noun. The possessor is, in a way, an adjective to the possessee. Fact checking this claim with WALS, we see that

  • 342 of 732 (46.7%) of surveyed Noun-Adjective languages favor Genitive-Noun,
  • another 342 of 732 (46.7%) of surveyed Noun-Adjective languages favor Noun-Genitive,
  • 232 of 318 (73.0%) of surveyed Adjective-Noun languages favor Genitive-Noun, and
  • 65 of 318 (20.4%) of surveyed Adjective-Noun languages favor Noun-Genitive.

Furthermore, most VO languages tend to favor Noun-Adjective + Noun-Genitive, and OV languages tend to favor either Noun-Adjective + Genitive-Noun or Adjective-Noun + Genitive-Noun (see: WALS Features 83A, 86A, and 87A combined).

Demonstrative & Article Location

If your language has demonstratives and articles, you can have some fun mixing up their locations. Of the 1225 surveyed languages that have demonstratives,

  • 542 have Demonstrative-Noun,
  • 562 have Noun-Demonstrative,
  • 9 have a demonstrative prefix, and
  • 28 have a demonstrative suffix.

When I compared this against order of Adjective and Noun, the demonstrative strongly correlated with the adjective's location. The idea here is to decide whether you like

  • "These dogs", or
  • "Dogs these", or
  • something else.

Relative Clauses

Most languages (70.3%) favor Noun-Relative clause. But there are a variety of ways that relative clauses can be formed. VO languages overwhelmingly favor Noun-Relative clause, and OV languages can go either way. Some example sentences are

  • Farmers who grow wheat...,
  • Wheat who grow farmers..., and
  • Farmers (who) wheat grow (who)... the who can go in either position.

Embedded Clauses & Complementizers

Finally, decide the position of embedded clauses and their complementizers. VO languages favor Comp. + Emb. Clause whereas OV languages are evenly split. The common routes are exemplified by the following sentences. Note that I have chosen SV for these examples.

  • I know that shepherds watch sheep.
  • I shepherds sheep watch that know.
  • I that shepherds sheep watch know.

I think you could have a lot of fun playing with these orders. Try them out and see what you like. Experiment with the aforementioned components of sentences until you are happy.

Putting it all together...

Once you've decided all the above, try taking a text in English then reordering it to obey your conlang's syntax. Omit words that do not exactly fit your conlang or replace them with approximates. While you do this, think of ways to "un-English-ify" the words and syntax. You could even explore how to vary syntax while maintaining grammatical correctness. Go wild and have fun!

In Ganzen und Großen...

I recognize this is a really, really long response, but I hope it is of value to you. I found myself in a similar situation asking how sentences should be formed. By looking into crosslinguistic data, various guiding trends appeared. Using this data has helped me more confidently decide my conlangs' syntaxes. Remember, since we're dealing with conlangs, you can make it into whatever you want! Just make sure to remain consistent and give logical reasons for why things are the way they are. As for evolving your conlang's syntax, I would recommend starting cave-man simple then slowly increasing complexity. Or heck, just come up with the modern conlang and figure out the protolanguage later! You could look into common etymological routes for tenses, aspects, moods, and other features of interest.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.