Conlang template

[Torco]

Glad I could help! Though I find it hard to share your enthusiasm about that questionnaire, as should be very evident from my complaints about it above.

Oh, the structure is baffling, I assumed it was random or responding to some arcane, scholarly reason: something like it being a tribute to the writer's old theses guide, who published a book that was just his lectures turned into chapters which were stuck together into a book in alphabetic order. But what I liked is that if one were to become stuck, one'd just pick a random thing and answer it. I'm not fully sold on general templates, and if I were to make consistent and finished grammars for my conlangs (as opposed to documenting them in, of all things, drive spreadsheets or Excel back in the day) I'd probably try to look at some language that's similar to mine and use that as template: natural grammars, to use a strange term, don't all have the same chapter names so why should constructed ones?

[Qwynegold]

Ergative alignment ↗Ergativity for Novices
In ergative (or ergative-absolutive) alignments S is marked the same way as O. If the language has case, then this is the absolutive case. A is marked differently, and for cased languages this would be the ergative case. If only word order is used for marking the arguments, then A and O must appear on different sides of the verb, while S must appear on the same side as O would, e.g. AVO and VS.

Ooh, my posts got a mention! (Which reminds me… I really do have to get around to writing the last chapter…)

It has been very helpful. *thumbs up*

Direct alignment
In direct alignment none of S, A and O are distinguished from one another, not even by word order. (So the language would have be either verb-initial or verb-final, or not have any default word order.) Unfortunately I have no information about this other than that what is what relies “entirely on context and common sense” — Wikipedia. This alignment is extremely rare, though it occurs in some languages with split ergativity. ***Link to split systems.

I’d say that Wikipedia’s description is a bit overblown. You can have direct alignment with respect to case-marking — in which case there are no noun cases. Or you can have direct alignment with respect to word order — in which case you have free word order. Only if you have both do you need to rely ‘entirely on context and common sense’. (Though even this extreme sense is still attested, in some topic-prominent languages.)

You could say that English is one: we have accusative case-marking for pronouns, but direct marking for nouns. Still, for some reason the term ‘split accusativity’ seems to be pretty much never used.

I'm thinking that if a language doesn't have case marking, or uses the same case for all of S, A, O; but uses word order to distinguish these arguments, then it's not direct. In this template I'm not assuming the conlanger has case, so it's important to take also word order into account when determining the conlang's alignment.

Austronesian alignment
Languages with Austronesian (or Philippine-type) alignment have both accusative and ergative alignments. In these languages verbs are marked with either agent trigger (actor focus) or patient trigger (undergoer focus).

Or a bunch of other triggers, effectively corresponding to applicatives. (Wikipedia lists locative and instrumental focus as well.)

Yeah, I chose to not mention those because it's not really relevant in this section.

{***Why would one choose one trigger over the other?}

As it happens, no-one’s come to a firm answer yet about this,

Ah, I've seen something like this being discussed here on ZBB. -_-

Thanks for all the answers! I'm making little changes here and there...

[Qwynegold]

The main reference grammar template that I’m aware of is the Lingua Questionnaire, an amazingly thorough set of questions covering pretty much every aspect of grammar.

man, this is cool enough that it might just make me get back into building conlangs.

Glad I could help! Though I find it hard to share your enthusiasm about that questionnaire, as should be very evident from my complaints about it above.

+1

[Vardelm]

...it's important to take also word order into account when determining the conlang's alignment.

True, but alignment isn't an all-or-nothing thing, of course. You can ergative marking and accusative syntax, for instance. Ever since I read the Typology of Ergativity by Mcgregor (it's getting harder to find a link to a PDF...), I have found it more useful to stop thinking in terms of "is this language ergative or not?" and instead think about "in what way is this language ergative?". I think the same thing holds for other alignments as well. So, with English, you could say it's mostly morphologically direct alignment (no marking) except for pronouns, which are accusative (as mentioned). Verbs agree with the subject, and so are also morphologically accusative. English is syntactically accusative (word order), and for the most part semantically accusative, except for a few ergative verbs like "bake". So, with all of that, you can say English "is accusative" just because of the preponderance of features, but that doesn't mean English is not morphologically direct for nouns.

[Qwynegold]

I wrote about syllable numbers a little while ago, after all. Now, in this text I'm talking about things that I don't actually know that well, and making claims that I can't back up. But what better way to get answers than by posting the wrong answers online.



1.2 Number of syllables per word root
Some languages have a very notable constraint on the number of syllables its word roots consist of. With word root here we mean the part of a word that does not have any inflection or derivational morphemes on it. This is not necessarily the dictionary form (lemma) of a word.

For example, in Swedish almost all verbs end with -a. This -a does not carry any grammatical meaning, but it's not part of the verb root. For example, the root of läsa (read) is läs-. One could argue that the -a is a verbalizer, and that is why it's not part of the root. Furthermore, this -a is deleted in compound words, c.f. läsförståelse (reading comprehension).

In some languages, the root itself may not be a permissible word or syllable. This can make it hard to count the syllables in a root. In such a case, it's up to you to decide on how to count root syllables. Different languages may have reasons to count differently, so no general rules can be given.

For many languages though, the number of syllables per root varies in such a way that it's not a noteworthy feature of that language. Therefore it is not that important for you to define your conlang's maximum number of syllables per root. As for the minimum number of syllables, most (perhaps all) natlangs have 1 as the minimum.

In this question we are dealing with your conlang's native words, not borrowed words. We are also dealing with root length in general. For example Mandarin is known for having monosyllabic roots, but even Mandarin has a few native roots consisting of two syllables. In this question, we would classify Mandarin as having a minimum syllable count of 1 and a maximum syllable count of 1 anyway, because the vast majority of roots are monosyllabic.

Put down a ✗ in both columns to mark your conlang's minumum and maximum number of syllables per word root.

Table with the heading "Min. number of syllables per word root", and rows with the following values:
1
1.5
2
More than 2
Undefined

Table with the heading "Max. number of syllables per word root", and rows with the following values:
1 (monosyllabic)
1.5 (sesquisyllabic)
2 (disyllabic)
3 (trisyllabic)
4
More than 4
Undefined

Monosyllabic languages (1 syllable)
Mandarin and Vietnamese are two well-known monosyllabic languages. (Their minimum and maximum number of syllables per root is 1.)

Sesquisyllabic languages (“1.5” syllables) ↗WP
Languages of this type have words that consist of full (normal) syllable, and a minor (reduced) syllable (which is the “half” syllable). The minor syllable has a simpler syllable structure than a full syllable, and not all the phonemes of the language may appear in a minor syllable. The details vary from language to language. Sesquisyllabic natlangs may also have words that consist of only one, full syllable. Therefore 1.5 syllables is marked red in the column for minimum root length.

All sesquisyllabic natlangs are spoken in Southeast Asia. This feature seems to originate in the Austroasiatic family, but it has spread to nearby language through language contact. Though on Wikipedia it is claimed that Proto-Tai and Old Chinese also had this feature. In any case, this is a notable Southeast Asian feature, so maximum length of 1.5 syllables has been marked as orange.

Polysyllabic languages (2 or more syllables)
Most languages belong to this group. Most (perhaps all) languages have 1 as the minimum syllable number. Therefore the options “2” and “more than 2” in the column for minimum number of syllables are marked as orange; we don't know if languages like this exist because of lack of data. We also don't know what are the longest roots some language has. 4 is probably very unusual, and more than four more unusual still. So at least the option “more than 4” is marked orange. The longer word roots a language allows, the simpler syllable structure it likely has.

The “undefined” option
If your conlang is neither monosyllabic or sesquisyllabic, then it can be difficult to define its maximum root length. You can use this option if you can't or don't want to bother defining this.

A few languages defy the notion of “syllable”, e.g. Berber languages and some languages on the northwestern coast of North America. These language may have very long consonant clusters, and even entirely vowelless words. If your conlang is of this type, then you may choose this option for both columns.

Suggested grammar sections: Introduction (of the whole conlang, the very first section), morphology, phonology, phonotactics, syllable structure.

1.2.1 Length of the complete word vs. root
Can inflections or derivational morphology make complete words in your conlang longer than the root? __________ (space to write your answer)

In Hungarian for instance, roots are monosyllabic, but words are typically way longer than that because of morphology.

Suggested grammar sections: Morphology, phonology, phonotactics, syllable structure.

1.2.2 Other notes about syllables
If there are any other things about root length in your conlang that you want to mention, you may do so below. For example, maybe your language has a general rule for root length, but this rule is broken under a specific circumstance.

If you have chosen the sesquisyllabic type, explain how this works in your conlang. Is the minor syllable the first or second, or either syllable? (In natlangs it's always the first syllable.) How are minor syllables different from full syllables? (Consider syllable structure, which phonemes may be present, tone, stress.)

Box to type in.

Suggested grammar sections: Morphology, phonology, phonotactics, syllable structure.

[Qwynegold]

...it's important to take also word order into account when determining the conlang's alignment.

True, but alignment isn't an all-or-nothing thing, of course. You can ergative marking and accusative syntax, for instance. Ever since I read the Typology of Ergativity by Mcgregor (it's getting harder to find a link to a PDF...), I have found it more useful to stop thinking in terms of "is this language ergative or not?" and instead think about "in what way is this language ergative?". I think the same thing holds for other alignments as well. So, with English, you could say it's mostly morphologically direct alignment (no marking) except for pronouns, which are accusative (as mentioned). Verbs agree with the subject, and so are also morphologically accusative. English is syntactically accusative (word order), and for the most part semantically accusative, except for a few ergative verbs like "bake". So, with all of that, you can say English "is accusative" just because of the preponderance of features, but that doesn't mean English is not morphologically direct for nouns.

I know. There's going to stuff about split alignments as well. And later on in this template you will be asked to define how the different verb arguments are going to be marked; through word order, case inflection, or particles, or some combination thereof.

[bradrn]

Looks good so far! (Though I have some comments, as usual.) One improvement you might make: a lot of languages have minimal word sizes, often described in terms of morae

As for the minimum number of syllables, most (perhaps all) natlangs have 1 as the minimum.

There’s a couple of languages where roots can consist of only one consonant. Generally this can happen in agglutinating languages where the various prefixes and suffixes fill out the rest of the syllable. (e.g. I’ve seen it in Georgian.) Another case is the Trans-New Guinea language Kalam: it isn’t very agglutinating, but it has a verb d- ‘get, take, hold’ which is phonologically attached to the next word (or any affixes it may have).

In this question we are dealing with your conlang's native words, not borrowed words. We are also dealing with root length in general. For example Mandarin is known for having monosyllabic roots, but even Mandarin has a few native roots consisting of two syllables. In this question, we would classify Mandarin as having a minimum syllable count of 1 and a maximum syllable count of 1 anyway, because the vast majority of roots are monosyllabic.

…

Monosyllabic languages (1 syllable)
Mandarin and Vietnamese are two well-known monosyllabic languages. (Their minimum and maximum number of syllables per root is 1.)

Huh? I was under the impression that most Mandarin roots are disyllabic. Not sure about Vietnamese, but I think it has quite a few disyllabic roots as well. (The writing systems don’t help, of course.)

A few languages defy the notion of “syllable”, e.g. Berber languages and some languages on the northwestern coast of North America. These language may have very long consonant clusters, and even entirely vowelless words. If your conlang is of this type, then you may choose this option for both columns.

My understanding is that they still have a concept of “syllable” which works fine — it’s just that they allow consonants as well as vowels to occur in the nucleus.

Sakao, on the other hand…

[Moose-tache]

Huh? I was under the impression that most Mandarin roots are disyllabic. Not sure about Vietnamese, but I think it has quite a few disyllabic roots as well. (The writing systems don’t help, of course.)

I think most words in Mandarin are disyllabic, but each syllable would come from a historically identifiable root. As for Vietnamese, it has the opposite problem. Inherited Austro-Asiatic roots are mostly (entirely?) disyllabic in Proto-Austro-Asiatic. But the initial syllable has been reduced to nothing in most Vietic languages. I don't think either language has an iron clad rule that every syllable has to be a transparently analyzable root, but that's how it works out in most cases.

[Qwynegold]

Looks good so far! (Though I have some comments, as usual.) One improvement you might make: a lot of languages have minimal word sizes, often described in terms of morae

Oh yeah. I don't know what to do with that. Is there actual information about this anywhere? Even so, I believe languages with mora counting can still have their words be divided into syllables.

As for the minimum number of syllables, most (perhaps all) natlangs have 1 as the minimum.

There’s a couple of languages where roots can consist of only one consonant. Generally this can happen in agglutinating languages where the various prefixes and suffixes fill out the rest of the syllable. (e.g. I’ve seen it in Georgian.) Another case is the Trans-New Guinea language Kalam: it isn’t very agglutinating, but it has a verb d- ‘get, take, hold’ which is phonologically attached to the next word (or any affixes it may have).

Hmm, this makes things complicated. Well, I did write this:

In some languages, the root itself may not be a permissible word or syllable. This can make it hard to count the syllables in a root. In such a case, it's up to you to decide on how to count root syllables.

If people have these template breaking things, they'll just have to write a note about it in the text box for miscellaneous notes.

Huh? I was under the impression that most Mandarin roots are disyllabic. Not sure about Vietnamese, but I think it has quite a few disyllabic roots as well. (The writing systems don’t help, of course.)

As Moose-tache said, most words in Mandarin are compounds of two morphemes. Now some people don't like to call them compounds, I'm not entirely sure why. Maybe because they think that a compound must consist of two complete words, and most of these morphemes do not occur alone so they can't be words.

A few languages defy the notion of “syllable”, e.g. Berber languages and some languages on the northwestern coast of North America. These language may have very long consonant clusters, and even entirely vowelless words. If your conlang is of this type, then you may choose this option for both columns.

My understanding is that they still have a concept of “syllable” which works fine — it’s just that they allow consonants as well as vowels to occur in the nucleus.

Sakao, on the other hand…

Wait, do the speakers of those languages think in terms of syllables? And how would you know which consonant is the nucleus and how many syllables there are among a long string of consonants?

[bradrn]

Looks good so far! (Though I have some comments, as usual.) One improvement you might make: a lot of languages have minimal word sizes, often described in terms of morae

Oh yeah. I don't know what to do with that. Is there actual information about this anywhere? Even so, I believe languages with mora counting can still have their words be divided into syllables.

For more information than you would ever want to know, read Gordon’s Syllable Weight: Phonetics, Phonology, Typology.

As for the minimum number of syllables, most (perhaps all) natlangs have 1 as the minimum.

There’s a couple of languages where roots can consist of only one consonant. Generally this can happen in agglutinating languages where the various prefixes and suffixes fill out the rest of the syllable. (e.g. I’ve seen it in Georgian.) Another case is the Trans-New Guinea language Kalam: it isn’t very agglutinating, but it has a verb d- ‘get, take, hold’ which is phonologically attached to the next word (or any affixes it may have).

Hmm, this makes things complicated. Well, I did write this:

In some languages, the root itself may not be a permissible word or syllable. This can make it hard to count the syllables in a root. In such a case, it's up to you to decide on how to count root syllables.

If people have these template breaking things, they'll just have to write a note about it in the text box for miscellaneous notes.

Ah, sorry, didn’t notice that.

Huh? I was under the impression that most Mandarin roots are disyllabic. Not sure about Vietnamese, but I think it has quite a few disyllabic roots as well. (The writing systems don’t help, of course.)

As Moose-tache said, most words in Mandarin are compounds of two morphemes. Now some people don't like to call them compounds, I'm not entirely sure why. Maybe because they think that a compound must consist of two complete words, and most of these morphemes do not occur alone so they can't be words.

Could you give me an example of a compound which isn’t made up of two complete words? I’m having trouble understanding this.

A few languages defy the notion of “syllable”, e.g. Berber languages and some languages on the northwestern coast of North America. These language may have very long consonant clusters, and even entirely vowelless words. If your conlang is of this type, then you may choose this option for both columns.

My understanding is that they still have a concept of “syllable” which works fine — it’s just that they allow consonants as well as vowels to occur in the nucleus.

Sakao, on the other hand…

Wait, do the speakers of those languages think in terms of syllables? And how would you know which consonant is the nucleus and how many syllables there are among a long string of consonants?

Firstly, what do you mean by ‘think in terms of syllables’? I’m not sure anyone thinks in terms of syllables as they speak.

And as for syllable structure, it depends. For Salishan languages (famous for words like [xɬpʼχʷɬtʰɬpʰɬːskʷʰt͡sʼ]), my understanding is that they still have rigid syllable structures, and words are still pronounced with clear syllables — it’s just that they allow consonants in the nucleus. (Wikipedia says that ‘In general, a syllable may be C̩, CF̩ (where F is a fricative), CV, or CVC’, with some unpredictability in the syllabification.) Sakao is a bit trickier — its syllables definitely has a nucleus (every syllable has a vowel), but there’s no clear way to determine where each syllable ends and begins, and some people have questioned whether the concept of a ‘syllable’ is even meaningful for Sakao.

[Nortaneous]

Some languages have a very notable constraint on the number of syllables its word roots consist of. With word root here we mean the part of a word that does not have any inflection or derivational morphemes on it. This is not necessarily the dictionary form (lemma) of a word.

...

Sesquisyllabic languages (“1.5” syllables) ↗WP
Languages of this type have words that consist of full (normal) syllable, and a minor (reduced) syllable (which is the “half” syllable). The minor syllable has a simpler syllable structure than a full syllable, and not all the phonemes of the language may appear in a minor syllable. The details vary from language to language. Sesquisyllabic natlangs may also have words that consist of only one, full syllable. Therefore 1.5 syllables is marked red in the column for minimum root length.

All sesquisyllabic natlangs are spoken in Southeast Asia. This feature seems to originate in the Austroasiatic family, but it has spread to nearby language through language contact. Though on Wikipedia it is claimed that Proto-Tai and Old Chinese also had this feature. In any case, this is a notable Southeast Asian feature, so maximum length of 1.5 syllables has been marked as orange.

This is not the right way to think about this.

Some languages have root shape constraints. These can be about syllables, but don't have to be. Take Nama:

Lexical root words consist of two or rarely three moras, in the form CVCV(C), CVV(C), or CVN(C). (The initial consonant is required.) The middle consonant may only be w r m n (w is b~p and r is d~t), while the final consonant (C) may only be p, s, ts. Each mora carries tone, but the second may only be high or medium, for six tone "melodies": HH, MH, LH, HM, MM, LM.

Also consider the vowel constraint in the more abstract reconstructions of PIE: the vowel of a root is generally *e. This is a root shape constraint, but not a root syllable constraint.

"Sesquisyllabism" is a stage in the common SEA areal progression of root shape constraints: the languages tend to develop from Austronesian-style disyllabism to reduced vowel contrasts in unstressed syllables to Cə "minor syllables" to CC initial clusters to complete monosyllabization to compound-induced polysyllabism. This is not unique to SEA - in some sense, Germanic is sesquisyllabic, but it had initial stress instead of final stress. (For monosyllabization, see e.g. German haben > ham.)

[Qwynegold]

I've resumed this project again, at least for a little bit.

I've added a note to my document to think about moras in addition to syllables, per bradrn's suggestion. About Nort's suggestion: I'm not going to deal with phonotactics properly, I only added that bit about syllables and word length.

Right now I've had inspiration to write about parts of speech, so I'll deal with that now so that at least I'll get something done.



Parts of speech
To some extent, it's possible to make different analyses of which parts of speech a given language has. This section deals with how you choose to formally classify the parts of speech in your conlang.

Open classes are those parts of speech that easily gain new words and closed classes are those in which new words are very seldom created. Open classes usually contains lots of words (thousands) while closed classes contain relatively few words. Though it's not unconceivable that there could be an open class where old words die out as quickly as new words are created, so the class stays small.

In the following table, mark with a ✗ whether your conlang has this part of speech as an open class or a closed class, or if it doesn't have that class at all. Even if you check “none” for a part of speech, it doesn't necessarily mean that the language doesn't have words with that kind of function. It just means that you don't formally recognize this as a part of speech in the language. For example, you could check “none” for numerals, because in your language numbers are classed either as nouns or as adjectives.

Some parts of speech here have subcategories. It's possible to have mark one of these as existing in your conlang without checking it's super- or subcategories. For example, you can check prepositions checking particles or adpositions. If preposition is the only kind of adposition in the language, it makes sense to only check prepositions and not adpositions. But if the language has both pre- and postpositions, one could either check both of those parts of speech, or only check adpositions. It depends on what you consider relevant to classify in the given language.

The last supercategory in this table, determiners, includes subcategories alredy listed elsewhere in the table. This is because there exists different ways to categorize parts of speech. You may delete the doubled subcategories articles, demonstratives and quantifiers where they are not needed.

Part of Speech	Open class	Closed class	None
Nouns		█	█
Verbs		█	█
‧ Coverbs?	█
‧ Preverbs?	█
Adjectives		█
‧ Quantifiers	█
Demonstratives	█
Numerals	█
Pronouns
Particles
‧ Adpositions	█
‧ ‧ Prepositions	█
‧ ‧ Circumpositions	█
‧ ‧ Postpositions	█
‧ Conjunctions	█
‧ ‧ Subjunctions	█
‧ Articles	█
‧ Discourse particles
Adverbs		█
Interjections		█
Classifiers (measure words)
Ideophones		█
Determiners	█
‧ Articles	█
‧ Demonstratives	█
‧ Quantifiers	█

Other parts of speech (please list all and give a short explanation of them):
___________________________________________________________



In that table yellow means a check on that box would be an unusual feature. Red means a check in that box is unnaturalistic. What's unusual is subjective. For example I've marked non-concatenative stuff, sesquisyllables and austronesian alignment as unusual features.

My questions now are: 1) How do I mark pronouns, discourse particles and classifiers? In Japanese pronouns are sometimes described as an open class. But that's the only such language I know. How rare is this really? Discourse particles and classifiers I have no idea about. The options here are mark open class as unusual, mark closed class as unusual, or mark neither as unusual.
2) Is there anything besides nouns and verbs where the lack of the class is unnaturalistic?
3) Are ideophones a kind of interjection, or should be kept separate?
4) Should I remove the whole determiners thing? That would make this whole thing less messy. And from what I understand, this is mostly a thing in English.
5) Should I list coverbs and preverbs as a special kind of part of speech? I'm not listing intransitives, transitives, auxiliary verbs etc.
6) Is there any other part of speech that should be added to this list?