Conlang template

[Ares Land]

I have my doubts about oligosynthesis mostly because I don't know if it's ever been proven for any language, worse, I'm not sure it can be proven.

Consider Nahuatl.

Nahuatl has quite a restricted phoneme inventory and a simple syllable structure. The number of possible syllables is quite low as it is. Besides, as is expected, vowel deletion, assimilation and other changes occur when morpheme combine. To make things worse, our sources on Classical Nahuatl dispensed with glottal stops, marking long vowels, or something just spelt wrong.
All of that means there are a lot of chance resemblances.

To make it even worse, Nahuatl relies heavily on metaphors. And those metaphors are quite alien to us. in atl, in tepetl or atepetl 'water-mountain' means 'city-state'. Huitzilopochtli is 'hummingbird-left' -- you'd never guess it's the name of a god of war.
These metaphors are rooted into a worldview that pretty much disappeared. (I believe the explanation is that the hummingbird alludes to a warrior and that the left side is associated with resurrection...)

I feel a lot of trepidation contradicting Benjamin Whorf, especially as a beginning student... But under these conditions, it seems very easy to see compounds that just aren't there.

That said, I don't claim expertise in any of this; and I don't anything about Caddoan or Kalam!
Of course, from a conlanging perspective, the idea of oligosynthesis is a very powerful one. (I would add, one that many of us use without meaning to, because who's got time to build 4,000 roots?)

I would say oligosynthesis doesn't really fit into the usual typological categories. You could have an isolating language with very few morphemes (as Kalam seems to be). Or say, a triconsonantal system relying on a reduced set of triconsonantal root and patterns, plus affixes.

[bradrn]

Well, I don’t know much about Nahuatl… I just know that it’s been claimed to be oligosynthetic, but not any of the details. So I’m not terribly surprised to hear that it isn’t ‘really’ oligosynthetic.

That said, I don't claim expertise in any of this; and I don't anything about Caddoan or Kalam!

For Caddoan, see that discussion I linked (which has a link to a lexicon). For Kalam, refer to Pawley’s paper ‘A language which defies description by ordinary means’ (which I believe you can get as a chapter from a larger book, the name of which I can’t quite remember at the moment).

I would say oligosynthesis doesn't really fit into the usual typological categories. You could have an isolating language with very few morphemes (as Kalam seems to be). Or say, a triconsonantal system relying on a reduced set of triconsonantal root and patterns, plus affixes.

Isn’t that just what I said already?

[Kuchigakatai]

Caddoan is very close to it (see this discussion), and I believe Nahuatl is too. Another example (which I actually discovered just a couple of days ago) is the Papuan language Kalam, which has sentences like these:

yp

me

wik

rub

d

get

ap

come

tan

ascend

d

get

ap

come

yap

descend

g-s<a>p

do-PRES.PROG-SG

He is massaging me.

(What’s happening here is that Kalam has a closed set of about 120 verbs, most of which are horrifically vague; e.g. d is ‘control, get, touch, handle, cease, complete…’, while tk- is ‘sever, ford, give birth, pack (for travel), cut a tattoo…’. Due to this, Kalam sentences tend to contain lots of long serial verb constructions to give context and clarification for any action that may happen.)

I feel that that example includes a certain confusion of diachrony vs. synchrony. Should d ap really be thought of as "get come" rather than a single morpheme dap? That example reminds me of various similar such words in Chinese and Burmese, e.g.:

七上八下
qī-shàng-bā-xià
seven-up-eight-down
'to be very disturbed/confused, be at sixes and sevens'

ကုန်တင်ကုန်ချမောင်း
kun tang kun hkya. maung:
/kòʊɴ-tɪ̀ɴ-kòʊɴ-tɕʰa̰-máʊɴ/
goods-load-goods-unload-drive
'construction crane' (lit. a goods-loading goods-unloading driven thing; I got this example from Comrie's The World's Major Languages)

I might be making too much of an analogy here, but the Kalam example looks awfully like it could be:

yp wik dap-tan-dap-yap gsap
me rub come-up-come-down is.doing.SG

Bringing in etymology is not always a good idea... Should the word "epenthesis" be similar analyzed with an ep- prefix? Maybe English speakers who know Greek do think of that as a prefix, but I suspect that for most people the stem is epenthe- ending in the suffix -sis (as in mitosis, miosis, catharsis, crisis, osmosis, synthesis, hypnosis, Genesis... maybe some people think of it as epenthes-is though).

[bradrn]

Caddoan is very close to it (see this discussion), and I believe Nahuatl is too. Another example (which I actually discovered just a couple of days ago) is the Papuan language Kalam, which has sentences like these:

yp

me

wik

rub

d

get

ap

come

tan

ascend

d

get

ap

come

yap

descend

g-s<a>p

do-PRES.PROG-SG

He is massaging me.

(What’s happening here is that Kalam has a closed set of about 120 verbs, most of which are horrifically vague; e.g. d is ‘control, get, touch, handle, cease, complete…’, while tk- is ‘sever, ford, give birth, pack (for travel), cut a tattoo…’. Due to this, Kalam sentences tend to contain lots of long serial verb constructions to give context and clarification for any action that may happen.)

I feel that that example includes a certain confusion of diachrony vs. synchrony. Should d ap really be thought of as "get come" rather than a single morpheme dap?

Yes, it definitely should — despite the orthography, d really is a full word (it has an epenthetic vowel), and I believe both d and ap may be used alone as verbs (though that usage is pragmatically infelicitous).

EDIT: I looked at the grammar; turns out d doesn’t have an epenthetic vowel and hence often isn’t a (phonological) word.

That example reminds me of various similar such words in Chinese and Burmese, e.g.:

七上八下
qī-shàng-bā-xià
seven-up-eight-down
'to be very disturbed/confused, be at sixes and sevens'

ကုန်တင်ကုန်ချမောင်း
kun tang kun hkya. maung:
/kòʊɴ-tɪ̀ɴ-kòʊɴ-tɕʰa̰-máʊɴ/
goods-load-goods-unload-drive
'construction crane' (lit. a goods-loading goods-unloading driven thing; I got this example from Comrie's The World's Major Languages)

Possibly…? There certainly is a commonality, in that these all seem like idiomatic expressions, though it’s hard to know whether the cases are exactly comparable. But I wouldn’t be surprised to hear that Kalam serial verbs are far more productive and common than those of Chinese and Burmese.

I might be making too much of an analogy here, but the Kalam example looks awfully like it could be:

yp wik dap-tan-dap-yap gsap
me rub come-up-come-down is.doing.SG

I don’t think so, though this does raise the thorny question of how exactly one defines a word. I don’t have access to a Kalam reference grammar (my sources are a few articles Pawley wrote), so I’m not entirely sure about this, but I do think that the Kalam sentence genuinely uses multiple verbs rather than a compound.

EDIT: Possibly some more examples along the same lines might help:

gos tmey nŋ-
thought bad perceive
‘dislike, hate something’

mnm ag ask ay-
talk say avoid stabilize
‘leave or avoid (a topic)’

ap tan ap yap g-p-ay tam
come ascend come descend do-HAB-3PL junction
‘crossroads’ (i.e. ‘junction where they come up and go down’; a nominalization)

ñag jw yok-
shoot withdraw displace
‘rout (the enemy) in a war’

pwŋy pag yk-
impinge disturb open
‘prise something open or free’

d nŋ-
touch perceive
‘feel something (by touching)’

kwy ap-INFL nŋ-
odour come perceive
‘smell something’

And also, now that I look at the paper again, Pawley does say: ‘A parsimonious [single] lexeme analysis can be made for certain serial verb strings … such cases are a small minority … Most can be derived from generative rules and so do not qualify as lexemes in a parsimonious grammar [although i]t is true that many strings of verb stems also happen to be familiar collocations’. So these clearly are not single words.

[Ares Land]

Isn’t that just what I said already?

Oh, sorry!

[bradrn]

I would say oligosynthesis doesn't really fit into the usual typological categories. You could have an isolating language with very few morphemes (as Kalam seems to be). Or say, a triconsonantal system relying on a reduced set of triconsonantal root and patterns, plus affixes.

Isn’t that just what I said already?

Oh, sorry!

Actually, oops, just realised this was my fault — I managed to completely forget that I didn’t say that at all, in fact I originally said the exact opposite:

On the other hand, oligosynthesis is clearly quite similar to those more traditional types, as it is also primarily a matter of how many morphemes a word contains.

So, given that I said that originally, and then I forgot I said that and said the exact opposite, I am clearly very confused about this subject

[Qwynegold]

I won't reply to the latest posts, but thank you all! I've rewritten my thing again, and I can finally present to first question. *sigh* I'm writing in OpenOffice Writer, and tables and formatting and such are hard to copy-paste here, so things in italics are just descriptions of what's in my document. I hope you can make sense of everything.

Regarding oligosynthesis though, I'm already ahead of you. Features marked with orange in this text are very unusual or somehow exceptional or noteworthy. You probably shouldn't select those on a whim. A conlang with several orange features is probably going to be pretty bad, unless you know what you're doing. Features marked with red are not attested in natlangs. These features should not be selected if you are going for a naturalistic conlang.



1 Level 1 questions
1.1 Morphological type
An overview of the traditional division of languages into morphological types is given below. But bear in mind that in this template, we have divided things up slightly differently. This is because no natlang is truly 100% of any category, and because some of these types are not mutually exclusive. For example, a language could be both agglutinative and fusional (e.g. Navajo).

In the following sections you are asked to grade things on a scale. Think of these numbers only as vague guidelines. It's impossible to actually measure how much of any type a given language is. As you develop your conlang, you might discover that it doesn't really correspond to the grades you had decided upon here. If so, stop and think whether you are happy with what the conlang has turned into. If yes, simply change your gradings. Only if the conlang is not pleasing to you, should you make changes to the conlang itself to make it better match your original design plans.

Isolating/analytic languages ↗WP
The terms isolating and analytic are often lumped together (see the Wikipedia link for further discussion). These languages have few morphemes per word; a perfectly isolating language would have one morpheme per word. In other words, isolating languages do not have much inflection. Instead, grammar is expressed through functions words and word-order.

Fusional languages ↗WP
In fusional languages one morpheme can code for several different grammatical categories at once. For example in English, the -s at the end of verbs codes for present tense and third person and singular all at once. Furthermore, in fusional languages grammatical categories may also be inseparable from the word root. Grammar may be expressed for example through grammatical tone or through ablaut. Words in fusional languages typically carry only one, or very few affixes simultaneously.

Oligosynthetic
Oligosynthetic languages are not actually attested among natlangs, so choose this option only if you are not aiming for a naturalistic language. In an oligosynthetic language all words are made up from a very small number of morphemes with very wide-ranging or vague meanings, which are combined together to form some form of compound words.

Agglutinative ↗WP
Agglutinative languages use inflection to express grammar. Typically, an affix in an agglutinative language codes for only one kind of grammatical meaning. In agglutinative languages words can carry several affixes at once.

Polysynthetic ↗WP, ↗Polysynthesis for Novices
Polysynthetic languages are like agglutinative languages, except more extreme. They tend to form very long words by stacking lots of affixes on the same word. One typical feature of polysynthetic languages is noun incorporation, the phenomenon in which a noun gets “rolled into” a verb structure.

Non-concatenative ↗WP
In non-concatenative languages grammar is expressed by modifying the word root. In Semitic languages a word typically consists of three consonants, and different vowels are placed at different places between the consonants when the word is inflected. This kind of language is called a tri-consonantal root language. Note that tri-consonantal root languages are very much associated with Semitic languages, and are not found in other families. Note however that there are also other ways to express non-concatenative inflection than through the Semitic setup (see the link above).

Ways of expressing grammar, and associated morphological types
Function words (prepositions, postpositions, conjunctions, particles, etc.) → isolating
Word order → isolating
Affixes (suffixes, prefixes, circumfixes) → agglutinative, polysynthetic, fusional
Infixes → non-concatenative
Noun incorporation → polysynthetic
Ablaut (changing the vowel in a word, e.g. goose-geese) → non-concatenative
Grammatical tone → non-concatenative
Initial mutation (modifying the first consonant of a word, like in Celtic languages) → non-concatenative
Transfixation (inserting different vowel patterns into tri-consonantal roots) → non-concatenative
Reduplication (full or partial reduplication) → non-concatenative
Truncation (the removal of a part of a word as a form of inflection) → non-concatenative

1.1.1 Degree of synthesis
Choose your conlang's degree of synthesis by putting a ✗ on the scale below. 1 means that the average number of morphemes per word is 1, or close to 1. This makes the language definitely analytic. 3 stands for a “reasonable” number of morphemes per word, making the language agglutinative, and 5 for a “large” number of morphemes per word, making the it polysynthetic. 2 and 4 are borderline cases. For example English could probably be a 2, because English uses very little affixing.

Scale with values 1 (analytic), 2, 3 (agglutinative), 4, 5 (polysynthetic).

1.1.2 Degree of exponence
This category corresponds to what is traditionally called fusion. It deals with how many grammatical categories morphemes express on average. 1 means that each morpheme typically only express one grammatical category. 3 means that the language is very fusional, with many morphemes expressing two or more grammatical categories at once. 2 is kind of middle of the road. If you have chosen to have an analytic language (degree of synthesis: 1), then you will likely have 1 here as well. Though one could imagine an analytic language with function words that express several different grammatical categories at once.

Scale with values 1, 2, 3.

1.1.3 Degree of concatenation
Decide to what degree you want morphemes in your conlang to be fused to the word root. 1 means no or close to no non-concatenative features. 2 means slightly non-concatenative (e.g. English with its ablaut in nouns and verbs and some noun-verb pairs only distinguished by stress as in ˈrecord – reˈcord). 3 means very much non-concatenative (e.g. Arabic). If you have chosen to have an analytic language (degree of synthesis: 1), then you should probably choose 1 here, because without inflection you can't have non-concatenative features.

Scale with values 1, 2, 3.

If you have chosen grade 2 or 3, describe in what way your conlang will be non-concatenative. Will it have infixes, ablaut, grammatical tone, initial mutation, reduplication, truncation? Will it be a triliteral (or some other number) root language? Note that this last option is very remarkable unless you are doing some sort of Semitic descendant language in a future or alt-history setting.

Small text-box in which to write things.

1.1.4 Degree of oligosynthesis
Decide if you want your conlang to be oligosynthetic. Some languages have a relatively small number or word roots, but an extensive set of derivational morphemes allowing them to derive lots of words. Nāhuatl and Blackfoot have been described as being oligosynthetic because of this reason, though nowadays the consensus is that there are no oligosynthetic natlangs. Grade 2 here could be used for languages that function similarly to Nāhuatl or Blackfoot.

Scale with values 1, 2, 3.

If you have chosen grade 2 or 3, describe in what way your conlang is oligosynthetic or oligosynthetic-like. If you also have chosen your conlang to be fusional (degree of exponence: 2-3), explain how you can combine these two features.

Small text-box in which to write things.



What do you think of this?

[Ares Land]

A minor remark: polysynthetic languages can certainly show a degree of fusion (Iroquoian languages, for instance, have combined subject-object prefixes, Seneca goes even further) - though as compared with IE, they are on the agglutinative end of the spectrum.

I thought some more about oligosynthesis... I'm not entirely sure the concept applies well to conlanging. A lexicon of 2,000 words is pretty respectable from an artlang; and most of these should have a proper etymology, and be related to each other. With 2,000 words as your target, you'll have a fairly reduced number of roots!
Or, in other words, conlangs tend to be oligosynthetic by nature, due to the finite nature of their creators )
Oh, and maybe compounding should be mentioned -- one notable thing about Nahuatl is how many roots that look 'basic' are actually compounds...

[Qwynegold]

A minor remark: polysynthetic languages can certainly show a degree of fusion (Iroquoian languages, for instance, have combined subject-object prefixes, Seneca goes even further) - though as compared with IE, they are on the agglutinative end of the spectrum.

Are you suggesting this should be mentioned at the overview (chapter 1.1)?

I thought some more about oligosynthesis... I'm not entirely sure the concept applies well to conlanging. A lexicon of 2,000 words is pretty respectable from an artlang; and most of these should have a proper etymology, and be related to each other. With 2,000 words as your target, you'll have a fairly reduced number of roots!
Or, in other words, conlangs tend to be oligosynthetic by nature, due to the finite nature of their creators )

But these conlangs are meant to grow their lexicon, right? I have a conlang that's meant to be oligosynthetic, so it has 41 roots. And Toki Pona is also oligosynthetic, right?

Oh, and maybe compounding should be mentioned -- one notable thing about Nahuatl is how many roots that look 'basic' are actually compounds...

Hmm, in what chapter should they be mentioned?

[Ares Land]

Are you suggesting this should be mentioned at the overview (chapter 1.1)?

Yes, I think so. Or maybe just defining polysynthesis as a high morpheme-to-word ratio? Your definition of polysynthesis is good, but I'm afraid it could be interpreted in too restrictive a fashion.

But these conlangs are meant to grow their lexicon, right? I have a conlang that's meant to be oligosynthetic, so it has 41 roots. And Toki Pona is also oligosynthetic, right?

Well, in universe, their lexicon is supposed to be larger... And you're right, some creators want to play with a very restrictive number of root... Yes, on second thought, forget about my comment!

Oh, and maybe compounding should be mentioned -- one notable thing about Nahuatl is how many roots that look 'basic' are actually compounds...

Hmm, in what chapter should they be mentioned?

1.1.4, Degree of oligosynthesis.

[Qwynegold]

Okay, I'll have to think about those things. Meanwhile, I have some new questions.

I'm focusing on just grammar in this conlang template, because phonology is not the part where I personally get stuck. But there is one phonology related thing I want to bring up: how many syllables word roots consist of. I know of monosyllabic languages, like Chinese and Vietnamese, and of Austroasiatic sesquisyllabic languages. But I wonder, are there any languages where roots are always two syllables, or three syllables, or more? And are there languages with roots of varying length, but where the mininum length is more than two? What about languages with a maximum length of two syllables, or three syllables, or more?

[Creyeditor]

Okay, I'll have to think about those things. Meanwhile, I have some new questions.

I'm focusing on just grammar in this conlang template, because phonology is not the part where I personally get stuck. But there is one phonology related thing I want to bring up: how many syllables word roots consist of. I know of monosyllabic languages, like Chinese and Vietnamese, and of Austroasiatic sesquisyllabic languages. But I wonder, are there any languages where roots are always two syllables, or three syllables, or more? And are there languages with roots of varying length, but where the mininum length is more than two? What about languages with a maximum length of two syllables, or three syllables, or more?

Indonesian is a language where people have argued that most native content word roots are minimally bisyllabic. I would guess there are many more.

[Moose-tache]

Native Old Japanese roots are nearly always two syllables. This is also true of the vast majority of words that can be traced to Proto-Tungusic. Native Korean roots are usually two syllables or show obvious signs of being two syllables in the past. But I am certain that none of these languages had hard rules precluding any roots from being monosyllabic.

[Chengjiang]

While they’re mostly still etymologically transparent, the majority of Mandarin content words are two syllables, and I could see the language becoming something like this in the future.

[bradrn]

Quite a lot of languages have a constraint where words must be at least two morae. Most commonly that equates to being minimally CVC/CVː/CVCV, though of course languages differ in the exact definition of a mora. (I recommend Gordon’s Syllable Weight: Phonetics, Phonology, Typology for much, much more information.)

[Richard W]

Quite a lot of languages have a constraint where words must be at least two morae. Most commonly that equates to being minimally CVC/CVː/CVCV, though of course languages differ in the exact definition of a mora. (I recommend Gordon’s Syllable Weight: Phonetics, Phonology, Typology for much, much more information.)

All words, or all content words? British English has a bimoraic constraint on content words, though Estuarine English looks as though it may lose it by dropping final glottal stops.

[Richard W]

Native Old Japanese roots are nearly always two syllables. This is also true of the vast majority of words that can be traced to Proto-Tungusic. Native Korean roots are usually two syllables or show obvious signs of being two syllables in the past. But I am certain that none of these languages had hard rules precluding any roots from being monosyllabic.

Proto-Uralic comes in this camp.

[bradrn]

Quite a lot of languages have a constraint where words must be at least two morae. Most commonly that equates to being minimally CVC/CVː/CVCV, though of course languages differ in the exact definition of a mora. (I recommend Gordon’s Syllable Weight: Phonetics, Phonology, Typology for much, much more information.)

All words, or all content words? British English has a bimoraic constraint on content words, though Estuarine English looks as though it may lose it by dropping final glottal stops.

All content words, I believe, though I wouldn’t rule out a language where this constraint applies to all words in general.

[Moose-tache]

Native Old Japanese roots are nearly always two syllables. This is also true of the vast majority of words that can be traced to Proto-Tungusic. Native Korean roots are usually two syllables or show obvious signs of being two syllables in the past. But I am certain that none of these languages had hard rules precluding any roots from being monosyllabic.

Proto-Uralic comes in this camp.

Oh, sorry, I didn't know we were including conlangs.

[MacAnDàil]

Following from the discussion about dissyllable predominance, the mention of Indonesian is unsurprising as is. Indeed, as I mention in my Master's thesis (page 96-97 https://www.academia.edu/40314155/LE_LE ... E_MALGACHE), 94% of Proto-Austronesian roots are dissyllabic. This Austronesian tendency stays to some extent in Réunion Creole words, especially in those of Malagasy origin.

There appears to be a certain geographic continuity among the language (families) mentioned here, but we can of course not conclude anything substantial from that.

@bradrn: Thank you for the mention of Gordon’s Syllable Weight: Phonetics, Phonology, Typology, I'll check it out when I can.