Rawàng Ata: quick feedback on a couple of diachronic syntax issues?

[Salmoneus]

For context, Rawàng Ata is my long-gestating conlang, to some extent inspired (rather unscientifically) by Austronesian languages. Its verb system and alignment have always been ridiculously weird - not actually an Austronesian system at all (in some ways related, in other ways almost the opposite). Over the years, i've tweaked some of the details, and I've also repeatedly struggled to produce a coherent diachronic justification of its weirdness (beyond "it was a long time ago, I wanted something weird, and this seemed intuitively weird for some reason"). Lately I've been thinking about it again.

I don't have a big proper write-up for you, but I wanted to run some developments past you and see what you thought about how plausible (and/or interesting) they were. For now, we're ignoring a lot of confusing additional bits and focusing purely on bivalent, simple, fully-verby verbs.


As Rawàng Ata's diachronics have undergone many revisions, I'll stay clear of 'actual' historical forms altogether for now, and I'll just use illustrative "English" (or mangled English) equivalents, or glosses where it seems necessary. This isn't to imply that there is an exact one-to-one semantic or syntactic correspondance to the English example - I'm just trying to illustrate accessibly the general principles.

Chip in with any comments or questions you may have!

-------------------------------

1. "Original" pronoun dropping/non-dropping

Long ago, the ancestor of RA, let's call it RA-1 had a straightforward SVO syntax. Verbal agreement marking was very limited - only subjects were marked, and most markings had either been lost already or would end up being lost in fairly short order (and won't be discussed further here). It also had optional topicalisation through fronting, and even some topics that weren't verbal arguments.

RA-1 had some compulsory pronouns. In fact, let's take a step back to RA-0. In RA-0, it was necessary to include a pronoun to refer to a core verbal argument when any of these were true:
a) the argument was 1st or 2nd person (I ate the biscuit; the dog bit you)
b) the argument was a 3rd person meriting particular respect, in which case honorific pronouns were used. (not "the dog bit the judge" but the dog bit his honour the judge
c) the argument was a 3rd person that had been topicalised (not "(as for) the sailor, the dog bit" or "(as for) the boy, ate the biscuit", but the sailor, the dog bit him and the boy, he ate the biscuit.

--------

Now, by RA-1, this had developed in three ways. Firstly, because rule b) was closely associated with politeness, it came to be generalised: whenever EITHER argument required an honorific pronoun, BOTH arguments used pronouns. Similarly, rule a) had often been ignored in informal contexts, and hence was likewise associated with politeness, and the generalisation of rule b) was extended to generalise rule a): whenever EITHER argument was 1st or 2nd person, BOTH arguments required a pronoun. Hence the dog he bit you, I ate it the biscuit and the dog he bit his honour the judge, alongside plain the dog bit the boy.

Secondly, some pronouns required only by topicalisation (i.e. not by the above politeness rules) were dropped. Specifically, a topic-agreement pronoun was dropped when all of these were true:
- the topic was the agent
- the agent was animate
- the patient was inanimate

Thus: the boy, bit the apple; but the apple, the boy bit it (topic is not agent), the apple, it choked the boy (topic/agent is not animate), and the boy, they bit the man (patient is not inanimate).

These restrictions can be explained by noting that an animate agent-topic acting on an inanimate patient is the 'default' situation, the most expected situation, and hence is less in need of marking - the roles can be assumed from context easily. Whenever this 'default' is departed from, however, RA-1 insists on a pronoun to reduce ambiguity (and reassure the listener that she's heard correctly, something unusual is happening).


Thirdly, and finally, a new politeness rule developed in RA-1: where a topic referred to a woman, a pronoun was required for politeness, and if there was none, because of the dropping rule above, a particular female pronoun had to be added (by repurposing a word meaning 'dear one') - that is, the girl, dear one bit the apple. However, where a pronoun was already present, it was not replaced - thus the girl, they bit the man. Unlike the original politeness rules that mandated pronouns for both arguments, this much milder rule had no broader consequences.

------------------

So to recap, in RA-1, pronouns are used (whether or not a coreferential noun is present):
- for both arguments, when either argument is 1st person, 2nd person, or an honorific 3rd person
- otherwise, when an argument has been topicalised, provided that the topicalised argument is not an animate male agent acting upon an inanimate patient.

Otherwise, pronouns are not used.

[personal pronouns, to be clear. Of course, other sorts of pronoun can also be used when required]
---------------------

I appreciate that this system is somewhat odd, but to me it seems intuitive, in that it's the result of several intuitive and I think cross-linguistically not uncommon principles: corefential pronouns for topics, pronoun dropping in the prototypical scenario, politeness triggered by the use of certain pronouns, and politeness requiring the use of pronouns (including where necessary honorific pronouns) for all parties.

The oddest thing might be that last principle, given that politeness is often said to involve avoiding pronouns. But it should be remembered that these 'pronouns' will often be honorifics (or the opposite). Hence, instead of "the boy bit the judge", it would be the boy, this lowly one, bit his honour the judge.


Does this make sense, do you think, both in terms of how I've explained it and in terms of what's going on?

[zompist]

Seems reasonable to me. It looks like a mix between French (for the topicalization) and East Asian languages (for the honorifics).

As you probably know, languages that use honorifics for pronouns tend to go kind of crazy with it. So I'd expect to see a lot of gradations. Perhaps the non-honorific cases dwindle over time.

[Salmoneus]

[going to have another go at this post]

[Salmoneus]

Seems reasonable to me. It looks like a mix between French (for the topicalization) and East Asian languages (for the honorifics).

As you probably know, languages that use honorifics for pronouns tend to go kind of crazy with it. So I'd expect to see a lot of gradations. Perhaps the non-honorific cases dwindle over time.

Thanks.

Actually, what ends up happening in Rawàng Ata is the opposite, but for the reasons you give. Using honorifics becomes a burden, so pains are taken to avoid it, including avoiding 1st and 2nd persons altogether (and using fictive 3rd persons). Of course, the array of fictive 3rd persons then becomes overly large as well...

[gestaltist]

I'll admit that second post was a tad confusing but I'm also far from an expert. Going one by one, I think the developments from RA-2 to RA-3 make sense. One thing I was wondering reading this: since instrumentals and agentives pattern together in RA-3 (same marking options, and they are the only two that have "MA" as the final particle), wouldn't it be likely for the agent of the agentive pattern to also be fronted? Especially that all other options have a somewhat less clear-cut agent.

BTW, I vaguely recall reading a PDF on Rawàng Ata. Am I dreaming or do you have a grammar available somewhere. It would've been a good while ago that I'd read it.

EDIT: I agree with zompist that the first post looks reasonable, so nothing to add here.

[Salmoneus]

Hmm. Maybe writing in a stream of consciousness about something complicated isn't a good idea after all... i'm going to have to take another run at that, aren't I?


On PDFs, btw: I have, in the past, written pdfs of different bits of the language, but I've never completed a grammar of the entire thing at one time.

[Salmoneus]

OK, I'm going to go a bit more slowly, and also completely ignore partitives and reciprocals for now for the sake of simplicity - I think I was trying to smash through too much too quickly in my summary. So, here's the next few points...



-------------

2. Outline of transition from RA-1 to RA-2
RA-1 developed into RA-2 through two significant changes (relevant to our topic here). One was an expanded role for topicalisation: all main verb clauses could now be regarded as possessing topics. The other was the fronting of verbs; the compulsory pronouns preceding and following verbs followed the verb and joined to it as bound morphemes.

Two clause types thus emerged. “Internal” clauses typically had the structure VAP (I’m avoiding terms like ‘subject’ and ‘object’ for simplicity, and sticking with ‘agent’ and ‘patient’), with either A or P being the topic. Where P was the topic, it could be promoted ahead of A, where the verb was single-marked and the meaning clear. Internal clauses could contain verbs that were double-marking (with honorific or 1st/2nd persons), single-marking (agent or patient), or zero marking in the specific case of a male animate agent topic acting upon an inanimate patient (which for simplicity we’ll call “default action”). “External” clauses typically had the structure TVAP, with a topic that was neither the agent nor the patient of the verb; external clauses also contained verbs that were zero-marked.


3. - the development of YA to RA-2
A key part of the syntax of RA is the particle we’ll call, for sake of argument, ‘YA’ (though this may never have been its exact form in any single stage of the language).

In RA-0, YA was a postposition with a wide range of uses, prototypically comitative, broadly similar to English ‘with’.

By the time of RA-1, YA was being used more broadly: it had an important genitive function, as well as a more specialised partitive function. It also had an instrumental function, marking the tools used to complete certain tasks.

In RA-2, YA’s role marking instruments was expanded to a role marking non-instigatory or non-volitional “pseudoagents”, where the topic was the true causer of the action. That is, rather than simply being used in statements of the kind “the man cut the bread WITH a knife”, it could also be used in statements of the kind, “the lord captured the castle WITH an army”, and “the man swept the floor WITH a servant”. In these situations, the causer was placed as an external topic, and the pseudoagent as an agent marked with YA. This structure was also found particularly in verbs of instigation of thought or perception – as, “the man [cause] made the woman [pseudoagent] see the monkey [patient]”.

A third kind of clause was also possible: one with a covert topic. This was an external clause, with the external topic left implicit – that is, where the topic is already established and need not be reiterated. These could be distinguished from agent-topic default action clauses by the YA marking of the “agent”. Thus, it was possible to distinguish:

CUT [knife YA] [bread] - “[they] cut the bread with the knife”
CUT [knife] [bread] - “the knife cut the bread”

and likewise:
SEE [woman YA] [monkey] - “the woman was shown the monkey [by them]”
SEE [woman] [monkey] - “the woman looked and noticed the monkey”


-------------


4. - the MA particle in RA-2
With all clauses having topics, but some topics being covert, with some verbs failing to indicate the topic either by being double-marking or by being zero-marking, and with limited gender or class agreement on verbs, and with a degree of flexibility in word order, by the time of RA-2 speakers found it convenient to make more explicit the emphasis of a clause. This was done not by explicit marking of topic, but by explicit marking of focus – the ‘new information’ being emphasised. While a specific particle was used in focusing verbs themselves, nouns were focused by means of a particle we’ll refer to as ‘MA’. MA probably originated in RA-0 as a simple, fairly bleached deictic of some sort - “cat MA” indicating “this cat” or “the cat here”. By RA-2, however, it was used as a focus marker.

However, the focus marker MA was not used where the particle YA (which inherently marked a non-topic) was already present.

Thus, of the agent and patient: one could be marked with YA and the other unmarked, in which case the marked argument was the focus; one could be marked with MA and the other unmarked, in which case the marked argument was the focus; or one could be marked with MA and one with YA, in which case the MA argument was the focus. In the cases of double-marked or zero-marked verbs where only one argument was marked with a particle, there was ambiguity between the unmarked argument being the topic and a covert external argument being the topic.
Fully agentive clauses therefore had the forms:
a) [topic] VERB agent MA patient; OR
b) [topic] VERB agent patient MA; OR
c) [topic] VERB-FOC agent patient
In the first case, the agent was the focus, while, in the absence of an explicit topic, the topic was either the patient or covert; in the second case, the patient is the focus, and the topic is either the agent or covert (or an explicit preposed topic); in the third case, the topic may be anything, and the verb is focused by its own means, which needn’t be discussed here. Fully agentive verbs could be double-marking (honorific), agent-marking, patient-marking, or zero marking (covert or external topic, or default action).

Meanwhile, pseudoagentive clauses instead had the forms:
d) [topic] VERB agent YA patient; OR
e) [topic] VERB agent YA patient MA; OR
f) [topic] VERB-FOC agent YA patient
...where in all three cases the topic is external or covert, but in the first case the agent is the focus, while in the second case the patient is the focus, and in the third case the verb focused by its own means. Pseudoagentive verbs were always zero-marking.


------------------------
-------------


So, how are we doing so far? Incomprehensible? Unrealistic? So plausible it's boring?

[zompist]

Are there many languages where focus is explicitly marked? Languages seem to be more eager to mark topics, perhaps because that's a more tractable problem... most of the sentence, ideally, is new information.

[Salmoneus]

Well, short answer yes, lots of languages have focus-marking (it's apparently widespread in Niger-Congo, also found in Somali, etc).

Longer answer: looking more closely, I'm not sure it works quite how I want it to in most languages.

[akam chinjir]

It doesn't have to be about new information, though. Some examples, with italics to mark (narrow) focus:

• They only criticised Sam's fur coat (and not, e.g., splash paint on it)
• They only criticised Sam's fur coat (and not, e.g., Mel's)
• They only criticised Sam's fur coat (and not, e.g., their leather shoes)

A fair number of adverbs and such can interact with focus in this way.

It can also play a role in parallelism/contrast:

• Sam brought the hummus and Mel brought the kimchi

(I think many languages would treat "Sam" and "Mel" as topics in this sort of construction, since they're the pegs that the sentence gets hung on, but you could also have them focused for contrast, which is how English intonation at least tends to treat them, I think.)

A cute example with both contrast and focus-sensitive "even":

• Read it, I haven't even taught it

A bit more relevant to Rawàng Ata: are you saying that all sentences will have a focused element? How would you, say, answer the question "What happened?" (I think answers to that question are usually taken to have focus on the whole sentence, which is to say that they lack narrow or constituent focus.) Or what about (what I think is the usual unmarked case) when the predicate as a whole is focus. E.g.:

• They only criticised Sam's fur coat (and not, e.g., sing the Internationale)

[Salmoneus]

A bit more relevant to Rawàng Ata: are you saying that all sentences will have a focused element? How would you, say, answer the question "What happened?" (I think answers to that question are usually taken to have focus on the whole sentence, which is to say that they lack narrow or constituent focus.) Or what about (what I think is the usual unmarked case) when the predicate as a whole is focus. E.g.:

• They only criticised Sam's fur coat (and not, e.g., sing the Internationale)

Well, to be clear, we're still dealing with a language in RA's past. The widespread use of focus particles will be temporary. Although 'modern' Rawàng Ata is also (/still) topic-prominent, so some of these issues will recur...

In answer to 'what happened?', you would say what had happened, with the same focus and topic as usual. After all, sentences like "they only criticised Sam's fur coat" are themselves valid answers to "what happened?" and can receive the same patterns of emphasis as on any other occasion.

Regarding the second case: I'm not sure. My instinct is that only one thing has to be chosen to be the focus - as in Somali, for instance - but I'd be OK with marking both a noun and a verb for focus. It's just marking two different arguments that I don't want to happen.



-------

Zompist: I've come across several people saying in passing that Quechua has explicit, morphological focus marking. Know anything about that?

[zompist]

Zompist: I've come across several people saying in passing that Quechua has explicit, morphological focus marking. Know anything about that?

Looking at a few papers... they're referring to the evidentials, such as mi (for direct knowledge) and si (for hearsay). The primary meaning is the evidentiality, but they can also be used for focus (or emphasis). The particular mix of uses seems to vary by dialect.

E.g., some data from Ecuadorian Quechua:

Ñuka-mi achku-ta kati-rka-ni.
I-evid dog-acc chase-past-1s
I chased a dog.

Ñuka achku-ta-mi kati-rka-ni.
I dog-acc-evid chase-past-1s
I chased a dog.

In both sentences -mi emphasizes the reliability of the statement; but which word it's applied to is the focus, like English stress.

[Salmoneus]

Thanks - that's a great idea I really should have thought of. I might very well have the focus particle descend from some original evidential, as that seems in keeping with my idea of the proto-language. If not, I'll have to use that elsewhere...