(Conlangs) Q&A Thread - Quick questions go here

[LinguoFranco]

What do you think of this vowel harmony system? Is it plausible?

It's /a e i ɤ o ɯ u/

/o u/ are rounded

/ɤ ɯ/ are unrounded

/a e i/ are neutral

[Arayaz]

What do you think of this vowel harmony system? Is it plausible.

It's /a e i ɤ o ɯ u/

/o u/ are rounded

/ɤ ɯ/ are unrounded

/a e i/ are neutral

I think it works. Just say that all back vowels in a word must agree in rounding.

[Creyeditor]

What do you think of this vowel harmony system? Is it plausible?

It's /a e i ɤ o ɯ u/

/o u/ are rounded

/ɤ ɯ/ are unrounded

/a e i/ are neutral

/a/ is usually unrounded/back in rounding/backness harmony systems and paired (or re-paired) with a back rounded or front unrounded vowel.

[Arayaz]

Does anyone have any advice about how to make a reference grammar of a conlang in the style of a real reference grammar? (Aside from reading a lot of real reference grammars.) How should it be organized? What should it include?

[Salmoneus]

What do you think of this vowel harmony system? Is it plausible?

It's /a e i ɤ o ɯ u/

/o u/ are rounded

/ɤ ɯ/ are unrounded

/a e i/ are neutral

/a/ is usually unrounded/back in rounding/backness harmony systems and paired (or re-paired) with a back rounded or front unrounded vowel.

Are there any other languages which only have rounding harmony in back vowels to compare it to?

[Salmoneus]

Does anyone have any advice about how to make a reference grammar of a conlang in the style of a real reference grammar? (Aside from reading a lot of real reference grammars.) How should it be organized? What should it include?

It should include whatever is necessary to fully describe the language, and organised in the best way to convey that information clearly.

I'm sorry that that's a non-specific answer, but there is no specific answer. One language will need a chapter one one thing, while another language will need a chapter on a different thing. One language may be best explained by treating two things as distinct phenomena, while another language might require you to treat the same two things simultaneously, because they are so interrelated in this language. One language will encourage you to describe X before Y, while another is really hard to describe without covering Y first and X only later. And so on.



---------

One way to look at things might be to think of an "ideal" grammar that explains everything completely, and work from there.

Such a grammar would:
- explain the context of the language
- explain the orthography being used in the grammar, and in other contexts
- explain the segmental phonemes of the language
- explain the phonotactics of the language
- explain any supersegmental phonemes of the language
- explain the intersegmental phonological processes of the language (harmony, sandhi, etc) within and between words
- explain the prosody of the language
- explain all the parts of speech in the language
- explain all possible morphology for each part of speech
- explain the permissible syntax of phrases for each part of speech
- explain the permissible syntax of every type of clause
- explain the permissible syntax of every form of conjunction of clauses
- explain the function of each syntactic permutation of phrases
- explain the function of each syntactic permutation of clauses
- explain the function of each syntactic conjunction of clauses
- explain the possible methods for conveying each possible perlocution and illocution, and the reasons why certain methods may be preferred in certain circumstances
- explain the principles of conversational language usage in the language, in varying registers
- explain the principles of prose composition in the language, in varying registers
- explain the principles of poetic composition in the language, in varying registers
- repeat all of the above for each dialect and sociolect
- explain the interactions of dialects and sociolects
- explain processes of code-switching, borrowing and calquing into and from the language into all other languages
- discuss all of the above diachronically, at each stage of the language's development since the beginning of time
- discuss trends in current language use and possible extrapolations into the future
- provide a corpus of exemplar texts in the language


In practice, no grammar ever does all of this, and rarely in this order. Not just because it would be hard, but because it is often inefficient - too much time would be spent repeating yourself or stating the obvious. So things are missed out or abbreviated, and things are moved around in order in a way that makes more sense for the specific language in question.

[VaptuantaDoi]

Take all of the following with a healthy grain of salt; I have no qualifications at all. And I've never written a complete reference grammar, but I've gotten about half-way through before.

Does anyone have any advice about how to make a reference grammar of a conlang in the style of a real reference grammar? (Aside from reading a lot of real reference grammars.)

Thomas Payne's Describing Morphosyntax is a guide intended for field linguists that works fairly well for conlangs too. There's also a much more informal guide mixed into Mark Donohue's Warembori grammar sketch.

How should it be organized?

Depends very much on the language. Generally in my experience there's two (good) ways to organise a grammar; either bottom-up (along the lines of word classes > morphology > NPs and VPs > clauses > complex clauses > further stuff) or top-down (e.g. clauses > NPs and VPs > morphology). I think synthetic languages suit the former and analytic languages the latter. And the top-down approach can be daunting – it feels very weird to start off with clause chaining constructions. It's also good to put a summary at the beginning so your readers know vaguely what they're looking at – basically just give an outline of basic syntax and morphology.

Phonology tends to be tacked on at the beginning. It too can be ordered either way - although personally it's far more readable to go bottom up (phonemes > allophones > phonotactics > prosody) rather than top-down.

On the other hand I've heard rumours that linguists have been known to order reference grammars alphabetically – that's if you're taking it to be a *reference* reference grammar which isn't meant to be read cover to cover.

I like to order things systematically, something like

2. Syntax
2.1. Basic clauses
2.1.3. Morphosyntactic alignment
2.1.3.4. Agency marking
etc.

That way you can easily refer to other sections. This is especially useful because languages don't lend themselves to neat ordering schemes – e.g. if you're explaining a morphophonological process you have to refer to morphology that won't be discussed for a while. I think it's impossible to describe any language entirely linearly.

What should it include?

As much as possible – the main things being phonology and morphosyntax. Phonology should cover the phonemic inventory, allophones, syllable and word structure, phonological processes, prosody, and if you want to get fancy things like consonant frequencies and historical phonology. Morphology is very language-specific; some languages don't really have any and you can cover everything under syntax (or you can lump it as "morphosyntax") but other languages have 33 verb slots and need a chapter just for that. In a synthetic language you should cover both grammatical morphology and derivational morphology for all relevant classes. It's a good idea to discuss word classes too at some point, which is a convenient time to list things like numbers and pronouns. Syntax should cover NP and VP (and any other phrase types) order, clausal order, morphosyntactic alignment, constructions like questions and imperatives, subordinate clauses, and devices like conjunction, SVCs, clause chaining and/or tail-head compounds etc. If you've got a true isolating language, lexicalised compounds still need to be discussed at some stage.

At the end you should always have a sample text – or at the beginning, or both. The lexicon can be at the end or as a separate book (depending on how long it is). Other things you might want to cover are semantics (no-one really knows how to incorporate this into a grammar though), pragmatics (ditto), discourse structure and other tenuous concepts. At the beginning some kind of introduction is generally expected.

But basically, just read or skim through lots of reference grammars, preferably modern ones which don't subscribe to some obscure crackpot linguistic theory. LangSci Press has a pretty good selection of well-written grammars, although they tend to be enormous (the Cayuga grammar is 1100 pages). Generally ignore anything that's not just "A (Reference) Grammar of X, (a X-an language)"; stick well away from "a Discourse-based Tagmemic Grammar of Tonajopoxtepec on Lexicostatistical Principles" or the like.

[Khemehekis]

What should it include?

You might want to check out my grammar of Kankonian, which runs to over 150 pages as a Corel WordPerfect document:

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

Ask yourself: how would Ruykkarraber tackle each of those example sentences? (Of course replacing vitzakhs (a Kankonian food) with some other food in your sentences, etc.)

[Creyeditor]

What do you think of this vowel harmony system? Is it plausible?

It's /a e i ɤ o ɯ u/

/o u/ are rounded

/ɤ ɯ/ are unrounded

/a e i/ are neutral

/a/ is usually unrounded/back in rounding/backness harmony systems and paired (or re-paired) with a back rounded or front unrounded vowel.

Are there any other languages which only have rounding harmony in back vowels to compare it to?

No, compared to languages with front rounded vowels and also backness harmony. Shibe and Shuluun Höh have systems with rounding harmony only in back vowels but they are pretty complicated and I don't really understand them.

[Nel Fie]

[...]
No, compared to languages with front rounded vowels and also backness harmony. Shibe and Shuluun Höh have systems with rounding harmony only in back vowels but they are pretty complicated and I don't really understand them.

Out of curiosity, I looked them up and found this paper which mentions both of those languages. Due to lack of any other hits, I'd infer this was your original source?

The Typology of Rounding Harmony in Phonetically Based Phonology, by Abigail R. Kaun (2004)

Most of it is steeped in Optimality Theory and tries to explain rounding harmony from that angle. Since I have no understanding of OT at this time, I don't understand the explanation, but an earlier part of the document also assays a plain English summary of how rounding harmony operates in Sibe and Shuluun Höh.

Though perhaps more usefully to the subject in general, it also offers a series of tables and text describing different types of rounding harmony found in natural languages. Might be a descent cheat-sheet for anyone interested in the matter.

[Creyeditor]

That's where I read about Shuluun Höh. I encountered some Sibe data in a more computational paper. But the original sources are cited in the source you give, IIRC.

[Nel Fie]

That's where I read about Shuluun Höh. I encountered some Sibe data in a more computational paper. But the original sources are cited in the source you give, IIRC.

Right. Based on a quick read-through of the above paper (not the original source), their rounding harmony operates as follows:

Sibe:

• Target must be [+back]
• Edit: Trigger may be of any placement
• For roots:
  • If trigger and target are of the same height, harmony always applies
  • If trigger and target are of different heights, harmony applies to high targets only
• For affixes:
  • Harmony applies to high targets only

Shuluun Höh:

• Target must be [+back]
• Edit: Trigger may be of any placement
• Harmony only applies if the target is non-high
• Harmony only applies if trigger and target are of the same height

Does that match/clarify what you read about them at the time? Or was it more complicated than this? Sadly the main cited document Tungusic Vowel Harmony: Description and Analysis by Li Bing from 1996 is not freely available as far as I can tell.

If the above is correct, then Shuluun Höh seems actually pretty simple, insofar that it can be reduced to "rounding harmony applies to back vowels of the same height as the trigger, except for [+high]".

Edit: The above was corrected from "rounding harmony applies within back vowel pairs of the same height except for [+high]"."

[Salmoneus]

Do the triggers also have to be back?

[Nel Fie]

Do the triggers also have to be back?

Right, the table didn't specify that so I forgot to include it. Apologies.

The adjoining text provides some clarification:

In Shuluun Hˆh (Svantesson, 1985), like other Mongolian languages, rounding
harmony is observed as long as the trigger and target agree in height and the trigger is
nonhigh. Shuluun Hˆh imposes the additional requirement that the target of harmony must be
[+back]. Rounding harmony in Sibe, a Tungusic language of China (Li 1996), imposes this
condition as well, and will be discussed in greater detail in Section 3.

The section 3.2 about Sibe in question states the following:

In Sibe, a southwest Tungusic language described in Li (1996), a rounding harmony
pattern similar to that observed in Type 6 is exhibited within roots: as with other Type 6
languages, cross-height harmony is tolerated only with a preferred (i.e., high) target. Nonhigh
vowels undergo rounding harmony as long as the trigger is also nonhigh, while high vowels
undergo harmony without regard to the height of the trigger.

Along with a few other notes here and there about the Types, that would suggest that no, the trigger does not have to be [+back]. It just has to agree in height for non-high targets, and for Sibe the trigger can be of any height for high targets.

[Ahzoh]

Voiced codas can lengthen vowels, but I wonder if long vowels can voice codas?

[Creyeditor]

@Nel Fie: I am still not sure if I understand the relation between the concrete data and the high-level abstract generalizations. But I can't think of many concrete questions. Going back to my answer to the original question: Does /a/ alternate in these languages or is it neutral?
@Sal: I think, you are right in implying that these are still not the perfect analogue of the proposed system.

[Creyeditor]

Voiced codas can lengthen vowels, but I wonder if long vowels can voice codas?

I searched P-Base and I got eight results for such processes. Here are the most convincing cases.

Code: Select all

Pech (Paya)	Rule	/k/ ~ [ɡ] / long X__	
Estonian	Rule	X → voiced / resonant__resonant, VV__resonant
Macuxi 	        Rule	X → voiced / {nasals, long vowels}__ (k → ɣ/ ɨː__)	
Muruwari	Rule	X → voiced / {long vowel}__	

[Salmoneus]

Voiced codas can lengthen vowels, but I wonder if long vowels can voice codas?

In my conlang, Middle Wenthish, a Germanic language, I voice consonants after long vowels - and, since this isn't something that comes naturally to me or that I've done in any other language, I presume I had a good reason for this based on something Germanic-y I read about. But I'm afraid I don't know what it was anymore!

The correlation of longer vowels before voiced consonants (not just codas) is apparently valid in English, Dutch, German, Norwegian, Swedish, Icelandic, French, Spanish, Italian, Russian, Lithuanian, Hindi, Assamese, Bengali, Telugu, Arabic, Japanese, and Korean, and with some exceptions in Welsh. The opposite correlation is allegedly not known of in any language.

And there are several studies out there showing that speakers of various languages take vowel length as a cue to interpret voicing - if you play people a longer vowel followed by a voiceless consonant, they'll often hear a voiced consonant. Which in theory SHOULD make it really easy for length to lead to voicing...

...but I'll admit I'm a bit concerned that there aren't a heap of obvious examples of this actually happening in any language!

[Ahzoh]

In Vrkhazhian the consonants /h j w/ elided in all places but word-initially (with some exceptions), which resulted in a lot of compensatory lengthening or vowel coalescence. The language has primary stress but also secondary and tertiary stress that are predictably placed every other syllable going leftward from primary stress. Thus trochaic foot (SU, SUSU, SUSUSU, etc.) Given some complex stress rules (if the penultimate syllable is light but the antepenult is heavy) sometimes the foot is shifted leftward one syllable (thus SUU, SUSUU, SUSUSUU).

Anyways this leads to issues in the paradigm that I must fix. So I'm wondering if these repair strategies make sense:

Spoiler:

Code: Select all

Predicative Active:
rabánni       > rabánni
nárabánni     > narbánni
tárabánni     > tarbánni
nayárabánni   > náyarbánni > nā́yarbánni 
tayárabánni   > táyarbánni > tā́yarbánni
rábadánni     > rabdánni

Attributive Active:
írrabánni     > írrabánni
innárabánni   > ínnarbánni
ittárabánni   > íttarbánni
ínnayárabánni > innáyarbánni > innā́yarbánni 
íttayárabánni > ittáyarbánni > ittā́yarbánni
irrábadánni   > írrabdánni

Predicative Applicative:
rabbánni       > rabbánni
nárabbánni     > nárabbánni   > nā́rabbánni
tárabbánni     > tárabbánni   > tā́rabbánni
nayárabbánni   > nayā́rabbánni > nī́rabbánni
tayárabbánni   > tayā́rabbánni > tī́rabbánni
rábbadánni     > rábbadánni

Attributive Applicative:
írrabbánni     > írrabbánni
innárabbánni   > innárabbánni   > innā́rabbánni
ittárabbánni   > ittárabbánni   > ittā́rabbánni
ínnayárabbánni > ínnayárabbánni > innī́rabbánni
íttayárabbánni > íttayárabbánni > ittī́rabbánni
irrábbadánni   > irrábbadánni

I had decided that the consonants will be kept if they are preceded by a long vowel or if they are geminated. Though I'm not sure I can justify <ayā> becoming <ī> when it would normally just become <ā>.
Part of my uncertainty is whether syllable weight should have an effect on secondary and tertiary stress.

[Nel Fie]

@Nel Fie: I am still not sure if I understand the relation between the concrete data and the high-level abstract generalizations. But I can't think of many concrete questions. Going back to my answer to the original question: Does /a/ alternate in these languages or is it neutral?
@Sal: I think, you are right in implying that these are still not the perfect analogue of the proposed system.

To be entirely honest, I don't know about /a/. Based on the table provided on page 30, I would infer that in the case of Sibe, /a/ can be a target for harmonisation and alternate with /ɔ/ in the case of roots and under the condition that the trigger is of the same height, but is fully neutral in the case of suffixes. I can't copy-paste the table due to formatting, but here are the relevant cases presented, using * to mark incorrect forms (in the original, correct forms are marked with the icon of a pointing hand):

ROOTS

*ɔmal
ɔmɔl

uva
*uvɔ

SUFFIXES

ɔs-maq
*ɔs-mɔq

suxu-maq
*suxu-mɔq


So, if this is correct, we'd have a situation where /a/ is sometimes neutral and other times not neutral within a single language - with grammar being a decisive factor rather than phonology.

However, Kaun notes elsewhere that the first of those examples for suffixes is made up because the original source doesn't provide an instance of a suffix containing /a/ preceded by a rounded vowel of the same height which could serve as trigger - so in effect, you could discount it for lack of valid data.

Ergo, we can only be decently sure that /a/ does alternate for roots, but we can't be certain if it's actually neutral for suffixes.

As for Shuluun Höh or other languages, I haven't found anything. The paper seems to build a lot of its case on Sibe specifically.