on A and B classes

[Qvaak]

I kind of promised to explain, what I/we think we know about these class A and class B things Mr. Peterson uses. I'm not sure when or how this information will be on the wiki, so meanwhile this gotta do.

These classes are simply declination patterns. There are no wider semantical or grammatical implications. If you meet an inanimate noun in its dictionary form (ie. in nominative case), you can't easily infer, what the accusative will be - the accusative will either get an /-e/ suffix or it will not. To inform us, whether or not the /-e/ will be added, David began using class A (/-e/ will not be added) and class B (/-e/ will be added) markings. This is pretty much the whole deal.

It seemed at first that this /-e/ suffix was just a phonotactical thing. If the regularily derived accusative ended in a consonant or consonant cluster that was impossible for the end of the word (agaist the rules of Dothraki phonotactics, that is) an /-e/ was usually added. The whole issue seemed to consider only few of the inanimate words, and it seemed the only truly undeterminable accusatives were on the words with stems ending in geminates, because there the geminate was sometimes just degraded to a short consonant and no /-e/ was added. Since then it has become more and more apparent, that there are half a dozen reasons to add an /-e/ suffix, and so it seems you often can't dependably determine, if an inanimate noun belongs to class A or class B.  It's best just to know.

Some unclear things:

• Are geminates the only case where class B would be required for phonotactical reasons, but class A is still sometimes used by simplifying the ending consontant?
• If the nominative of an inanimate noun ends in consonant, is it always class A noun?
• It does not make sense (to me) to use these classes with animate nouns, but verb past singulars work so similar to noun accusatives, that this kind of classification might be an useful tool for them too. But they seem so regular, using /-e/ always when phonotactics demand and never otherwise, that perhaps there is no need?
• As far as we know, we have met precisely one inanimate noun with vowel ending stem, zhalia. Are all inanimate nouns with vowel ending stems like this: extra vowel in nominative and class B?
• What are all the reasons for nouns to be on class B? There is some symmetry/rhythm stuff, some homonymity dodging, some phonotactics... what exactly?

When we first came up with these (and some other) irregularities, we decided to keep the techical stuff at minimum. As the accusative (and in verbs the past singular) is the only hard to determine case, we decided to simply list it as a supporting declination instead of some techical coding. When David introduced these class markings, we thought we'd just keep the vocab as it was. It might be a time for a change, either to David's marking system, or to double system, with both support declinations and David's markings.

[Hrakkar]

Thank you, Qvaak, for putting out this post. I don't know if you saw it, but I tried to explain class A and B in the dictionary front matter, and 'missed significantly'. The next time I update it, I will instead put a link to this thread.

In the meantime, I do not have much in the way of comments for this matter, as you are way ahead of me on understanding it.

I think the idea of identifying type A and B is useful. I think showing the accusative or past tense form is also useful. Both would be OK, as long as they do not become confusing.

[ingsve]

I don't really think the class A and B are that useful as long as we also have the accusative and past tense marked in cases where it matters. Adding a whole new coding to all nouns in the vocabulary (that in most cases doesn't change anything) is just likely to confuse people which is why we opted to not mark it in the first place. I don't think anything has changed since then.

[Hrakkar]

I have started to indicate A and B classes in the most recent dictionary. This is partly due to the trouble it takes to add new fields to dictionary entries. But making these stem forms clear to the user is worth the effort wt will take to do it.

[Najahho]

I would believe the best option is to do as the Latin dictionaries. In Latin you mention a verb also mentioning the genitive, so for example, campus, -i and you know it's gen. campi and the rest you can get from just that. So similarly, I think you could avoid going into the whole "class B" and "class A" if you just add the accusative. For example sondra, sondre; mawizzi, mawizze; jelli, jel, and so on.

I think the confusing part you mention is like why mawizzi > mawizze but jelli > jel, my guess would be that "mawizzi" is a derivate word, like the root probably means something else, but JEL could be the root in itself and "jelli" the noun derived from the root. Of course I have no way of proving this but it's my guess.

You could likewise use the genitive for animates, so you have rizh, rizhi but ko, koes.

[Hrakkar]

Niqqo, your last comment is a very interesting idea!

[Najahho]

Niqqo, your last comment is a very interesting idea!

Hehe Thank you for that! 

I also wrote to you about the zhalia > zhalie

[Qvaak]

I would believe the best option is to do as the Latin dictionaries. In Latin you mention a verb also mentioning the genitive, so for example, campus, -i and you know it's gen. campi and the rest you can get from just that. So similarly, I think you could avoid going into the whole "class B" and "class A" if you just add the accusative. For example sondra, sondre; mawizzi, mawizze; jelli, jel, and so on.

Isn't that almost exactly what we already do in the wiki vocabulary?

You could likewise use the genitive for animates, so you have rizh, rizhi but ko, koes.

Genitive of ko is kosi as far as I know. I think it's better to have all nouns with the same supporting declination. Using genitive for animates and accusative for inanimates sounds a bit confusing. Even though there are two declination patterns for animates, the mechanics are easy and predictable. The only irregularity we know of - mai, lai -> mayes, layes - actually manifests in accusative, and in these cases the wiki vocab already lists the accusative.

[Najahho]

Yes, I got confused, it was rather late at night. I meant ko, kosi. But if any of what I've said is already being used then never mind my comment, heh. I don't consider mai > mayes such a problem, you can't have "maies" in Dothraki if I'm not mistaken.

In any case, yes, animates are too regular and easy to predict to be taught in different paradigms.

[ingsve]

I would believe the best option is to do as the Latin dictionaries. In Latin you mention a verb also mentioning the genitive, so for example, campus, -i and you know it's gen. campi and the rest you can get from just that. So similarly, I think you could avoid going into the whole "class B" and "class A" if you just add the accusative. For example sondra, sondre; mawizzi, mawizze; jelli, jel, and so on.

I think the confusing part you mention is like why mawizzi > mawizze but jelli > jel, my guess would be that "mawizzi" is a derivate word, like the root probably means something else, but JEL could be the root in itself and "jelli" the noun derived from the root. Of course I have no way of proving this but it's my guess.

You could likewise use the genitive for animates, so you have rizh, rizhi but ko, koes.

The thing with words like jelli -> jel, tolorro -> tolor, etc. is that they are irregular nouns while mawizzi is regular. It's just one of those things you get in languages, that some words just don't follow the normal rules that other words do. If you create a language where everything is neat and regular it becomes less naturalistic and Dothraki is suppose to be a naturalistic conlang.

[Najahho]

I bet there's a root explanation to that.

[Qvaak]

I bet there's a root explanation to that.

Aye. Probable. Irregularities usually have their background in the history of the words. Might easily be several different explanations depending on the word. Might even be some explanations that are still to be locked in place.

[Qvaak]

I said:

It does not make sense (to me) to use these classes with animate nouns, but verb past singulars work so similar to noun accusatives, that this kind of classification might be an useful tool for them too. But they seem so regular, using /-e/ always when phonotactics demand and never otherwise, that perhaps there is no need?

After the discussion about hethke and hethkat on the blog, it's quite obvious that the A/B distinction is relevant with verbs too. Not that this makes much difference for the vocab - we already list the past singulars of (suspect) verbs.

[Hrakkar]

So, is a general pattern of two classes of declension emerging here, that crosses word function lines? Although most conlangs are pretty 'regular' by natural language standards, David has been pretty clever at 'designing in' some irregular patterns that add great interest to the language. Still though, the whole A/B class thing is still 'clear as mud' to me. As if it was meant for me to never understand it

[Najahho]

So, is a general pattern of two classes of declension emerging here, that crosses word function lines? Although most conlangs are pretty 'regular' by natural language standards, David has been pretty clever at 'designing in' some irregular patterns that add great interest to the language. Still though, the whole A/B class thing is still 'clear as mud' to me. As if it was meant for me to never understand it

What's to understand? I think it's pretty clear.