The CBB

Hello everyone, sorry if this has been posted before, I'm not very familiar with the search function or the boards here so I don't know if I'm repeating a common question:

What are some common pitfalls when documenting a conlang, especially its grammar?

Something like common mistakes made, by beginners or more experienced conlangers, maybe not including enough examples, not elaborating enough on some points, not properly defining terms?

I have been doing conlanging since 2015 but I still often feel very clueless when it comes to writing out grammar articles, unsure what to include, how to describe it etc. and some friends have pointed out before that my documentation could be better so I wanted to ask if there are some good things to keep in mind. I have a thread on sections in a grammar bookmarked, so this is asking more about general grammar writing rather than structuring.

Unfortunately, I don't think it's necessarily something where specific advice can be given.

The things you probably have to do are:
- explain everything your readers don't already know
- don't explain things your readers already know
- explain things in the correct order, explaining each thing in a way that doesn't rely on already knowing or understanding something that you haven't explained yet
- include enough examples, but not too many
- use standard terminology where possible, except where doing so would be misleading

You may ask: "but how is that possible!? How can I possibly write something that is neither longwinded nor overly cursory? And how can I explain things in order, when every topic seems to rely on other topics so that there is no clear starting point? And how can I stick to standard terminology designed for use in describing SAE languages, when my language is not SAE and every term I use seems misleading, but not using the terms is longwinded and requires the reader to learn a bunch of ad hoc jargon?"... and the answer is, it's not possible.

All grammars are badly written. It's probably impossible to write a good one. That doesn't mean you shouldn't try - do try to be as logical in your order as possible, do try to be clear in your terms, do try to give examples, but also do try not to be confusingly longwinded and repetitive - but it does maybe mean you shouldn't be too upset if you think you haven't succeeded: nobody so far has.

Thanks!

Personally, I'm chiefly concerned with writing a grammar that will help me recall the details of my conlangs. I'm failing even at that. Writing grammars is damn hard.

I enjoy examples with lots of examples and explanations that cover details of usage, e.g. when exactly do you use singular forms or accusative forms or present tense. This is hard for me and not a lot of conlang grammars have it. Also interesting example sentences are a plus.

Pitfall: Caring too much about grammar.
For conlangs that have less or almost no grammar, it is pointless to write a grammar book. If I were to document my conlang's grammar, I'd only end up with 2 pages, including a table of affixes. Instead, I choose to format it as a textbook, with more opportunities to show sentences that go beyond grammar (the most difficult facet in any language-learning process imo).

The conlanger should suit the style of documentation to the type/characteristics of the conlang. Sometimes, there is no need for extensive explanations like on wikipedia if they do not describe your conlang well enough.

I think a lot of documentation pitfalls can be summed up as "explaining form but not function". Conlangers explaining the morphological process behind how to form the dative, but leaving it up to the reader to jump to the conclusion that the dative is used for nominal predicates. Or explaining how serial verbs work, but leaving it up to the reader to jump to the conclusion that serial verbs are used to express object complements. (The latter actually happened a while back.)

In general, I think Reyzadren there illustrates perhaps the biggest pitfall: thinking that your conlang has "less grammar", or "almost no grammar". Languages mostly have around the same amount of grammar, because they're mostly equally complex*. When a conlanger fails to describe an area of grammar, this generally means that they're just assuming that things work as they do in their own native language (or, sometimes, in accordance with some sort of native-language-derived objective "logic", which of course is neither logical nor objective) - which both makes the conlang derivative, and makes the grammar confusing for anyone who doesn't share the creator's assumptions. This is primarily a syntax problem: people realise that they need to describe their morphology explicitly, but kind of forget that most of grammar is actually syntax.

So: don't overlook syntax!


*within some narrow band of complexity, and leaving aside creoles/pidgins/etc and mixed languages and perhaps some highly-complicated languages spoken by small and isolated communities.

Khemehekis wrote: ↑02 Apr 2021 08:26 I think a lot of documentation pitfalls can be summed up as "explaining form but not function". Conlangers explaining the morphological process behind how to form the dative, but leaving it up to the reader to jump to the conclusion that the dative is used for nominal predicates. Or explaining how serial verbs work, but leaving it up to the reader to jump to the conclusion that serial verbs are used to express object complements. (The latter actually happened a while back.)

I definitely agree. On the other hand, sometimes people describe isolated forms and allophony, but not their interaction. Simplified example: If the syllable structure is stricly CV, how do V suffixes actually work?

Creyeditor wrote: ↑02 Apr 2021 18:55 I definitely agree. On the other hand, sometimes people describe isolated forms and allophony, but not their interaction. Simplified example: If the syllable structure is stricly CV, how do V suffixes actually work?

Oh, that too!

http://dedalvs.com/megdevi.html

Here, David J. Peterson explains the reasons his first conlang failed. He notes here:

"The word "phonology" is in scare quotes because, properly speaking, Megdevi doesn't have a phonology. Instead, it has a list of invariable sounds. In this way, it's like Esperanto, which also lacks a phonology in its theoretical form (in practice, I'm sure there's phonological variation—especially for the L1 speakers out there). So if you had an Esperanto word like, oh, I don't know, taksii, meaning "to take a taxi", the final vowel isn't long: It's two discrete occurrences of the vowel i, the first one being stressed. And let's say you wanted a word for a "wheel trampoline" (whatever that is), you could create the word radtrampolino, where you would have to pronounce a voiced [d] followed immediately by a voiceless [t]. Try it; it ain't easy. Similar things happened with Megdevi. So, for example, taking the triconsonantal root b-t-dz for "lizard", you can get bitdza, a word meaning "lizard-colored". I'd say that cluster is nearly impossible to pronounce. Nearly.

. . .

What's surprising is that all of these are phonemes. That's 43 consonant phonemes in total, unless I miscounted. And further, there was no allophony. So in all situations, no matter what happened, an /s/ sounded like [s]; /t/ sounded like [t], etc. This might be all right if the max syllable of the language was CV, but if two consonant come into contact with each other, things inevitably happen.

And, just to clarify, a triconsonantal root was composed of three of the consonants above. They could be any three, including duplicates. There were no constraints on what could constitute a root. When deriving words, for those who aren't familiar with Arabic or another Semitic language, vowels were put in and around the consonants in various ways, but the three consonants of the root had to appear in the same order (I'll talk more about this later)."

His examples are pretty rubbish, though. Lots of languages have vowels in hiatus (including reduplicated vowels), and lots of languages allow clusters including voiced and unvoiced stops together. Even as a monolingual English speaker, I have no difficulty pronouncing his examples!

[I would struggle with mid-cluster devoicing if the cluster is [can't remember the word for 'in the same syllable']. But with the syllable break, a word like 'radtrampolino' is no trouble at all]

Salmoneus wrote: ↑03 Apr 2021 12:41 His examples are pretty rubbish, though. Lots of languages have vowels in hiatus (including reduplicated vowels), and lots of languages allow clusters including voiced and unvoiced stops together. Even as a monolingual English speaker, I have no difficulty pronouncing his examples!

[I would struggle with mid-cluster devoicing if the cluster is [can't remember the word for 'in the same syllable']. But with the syllable break, a word like 'radtrampolino' is no trouble at all]

Well, I agree with you about his specific examples. I have no trouble with taksii, scii, and the like (even English has the Hawaiian borrowing aalii!)

Pronouncing [np] without it sandhizing into [mp], on the other hand -- now, that's more difficult.

Salmoneus wrote: ↑03 Apr 2021 12:41 His examples are pretty rubbish, though. Lots of languages have vowels in hiatus (including reduplicated vowels), and lots of languages allow clusters including voiced and unvoiced stops together. Even as a monolingual English speaker, I have no difficulty pronouncing his examples!

[I would struggle with mid-cluster devoicing if the cluster is [can't remember the word for 'in the same syllable']. But with the syllable break, a word like 'radtrampolino' is no trouble at all]

"tautosyllabic".

Anyway, I agree.

I just searched for the sequence "dtr" in my Kankonian spreadsheet to find words like "radtrampolino". There are about a dozen:

adtrai: to round (a number)
adtrayaspayo: to mind-rape
adtrayor: to predicate
adtreil: to septuple
adtreilathe: to heptalize; heptalization (refers to the heptal numerus found on Javarti nouns and pronouns)
adtre*ia: to nominalize; nominalization
adtrisharab: at worst
adtrivenit: at best
fuedtreveniti: all-conference
lemadtrank: impossible to guess
meadtreshos: anxiolytic
tandtrenki: prepaid

And the real kicker: when I searched the English column for "dtr", it turned out that *English* had two common words with -ndtr- (one of which is often written with a space). Can anyone think of what they are?

soundtrack

DesEsseintes wrote: ↑03 Apr 2021 16:45 soundtrack

Good! You got one!

Salmoneus wrote: ↑03 Apr 2021 12:41 His examples are pretty rubbish, though. Lots of languages have vowels in hiatus (including reduplicated vowels), and lots of languages allow clusters including voiced and unvoiced stops together. Even as a monolingual English speaker, I have no difficulty pronouncing his examples!

[I would struggle with mid-cluster devoicing if the cluster is [can't remember the word for 'in the same syllable']. But with the syllable break, a word like 'radtrampolino' is no trouble at all]

I think he meant simply that his examples can be unstable over time. I mean, if I'm thinking about allophony, things like /‍ˈi.i‍/ → [‍ˈiː‍] or /‍ˈi.i‍/ → [‍ˈiʔi‍] or /‍dt‍/ → [‍tt‍] are among the most obvious possibilities; not inevitable, but worth considering.

As far as English goes, here I am saying "orange juice" with /nd͡ʒ.d͡ʒ/, no problem at all, while a lot of other people say only /n.d͡ʒ/. Pronouncing /nd͡ʒ.d͡ʒ/ isn't necessarily difficult, least of all for a native English speaker, but euphonic processes are at work nonetheless.

Sure, allophony (or diachronic sound change) CAN happen in scenarios like that, and often does. But he talks about these sequences as things that arise only when you "don't have a phonology", and that in some cases are "nearly impossible" to pronounce. In reality, these sequences may be more difficult to pronounce than many sequences - enough that they wouldn't occur in most languages - but are a long, long way from the boundaries of what's impossible to pronounce, and do occur in perfectly natural languages, not just Esperanto.

I think there's a particular trap here for English speakers, who may readily declare 'radtrampolino' impossible to pronounce, yet who merrily mutter their ways through "spinster", "minstrel", "spritz", "geoid", "strength", "alstromoerias", "steadfast", "paintstick", "meanwhilst" and "bloodthist" and so on...

He was incorrect to portray the clusters as "nearly impossible to pronounce". The trap for English speakers, though, is probably assuming that clusters like the ones you've listed are more common crosslinguistically than they truly are.

Word of the day: Alstroemeria

Oh man, Kankonian just *must* acquire a word for "alstroemeria"!

EDIT: "Angstridden" is a good one too.