Detail #232: Reciprocals vs Reflexives

Reciprocals (they saw each other), and reflexives (they saw themselves) are a couple of fairly similar meanings that might not even be distinguished in any systematic manner in a language. However, we could also imagine them being distinguished in haphazard manners.

Let us consider the actions described -

they verbed each other vs. they verbed themselves

Somehow, to me at least they verbed each other seems like a single action carried out by a group, while they verbed themselves seems like lots of small actions carried out by multiple people. This might of course depend on the nature of the verb - if the verb means wash your teeth, then clearly the latter implies lots of people independently doing things, whereas if cause to lose a war is the verb involved, then the communal action interpretation might be more close at hand.

So, why not have plural congruence in the case of the separate, independent actions and singular congruence in the other case, and slowly make this also turn into plural congruence with reflexives implies reciprocality, lack thereof implies reflexivity. (It could just as well fall out the other way, but the basic principle is what I'm getting at – of course, the distinction between communal action and individuated action could also be present without influencing reflexives vs. reciprocals at all.)

We could of course take that and go in the direction of ... inverse systems. Some verbs prefer a reciprocal interpretation, some verbs prefer a reflexive one, and the inverse marker turns the interpretation around?

But that's not particularly new an idea for this blog, even though it's not been mentioned before iirc. It skates close to exoreflexive and endoreflexive verbs.

A new idea would be this, though: take the same pronoun for reciprocality and for reflexivity. Thus "them|our|yourselves" and "each other" are conflated. Now, with the reciprocal meaning, the pronoun is marked as the subject, and the noun phrase that normally would have been the subject is the object:

theyselves helped them

and the regular reflexive meaning would be

they helped themselves

This is probably typologically unattested, and maybe even breaks some universals. I have a sort of inclination to think that if the language lacks a proper subject, though, it might be more likely. Such languages seem to be syntactically ergative relatively often, so there you have an idea to play with.

Detail #231: A Noun Complex

So, let's consider a language that has fairly morphologically simple nouns: maybe they mark one or two things - but preferrably at most one of those things at a time.

But the noun phrase has a thing going for it that is more complicated. The article. The article bears the brunt of morphological marking, and this to an extent that far surpasses German.

First, we'll posit a historical origin: nouns that have been grammaticalized. So, instead of "actor" vs "actress", you have "boy actor" vs "girl actor". Thus, what now are articles, historically have been nouns. There is a fair share of such former nouns, and the noun classes can be somewhat overlapping.

Possessors of the noun have a congruence marker in the article. Certain nominal possessors may be incorporated into the article, but some will just have a possessive pronominal suffix on the article. Certain adjectives can also be incorporated, but only one at a time.

A number of morphemes corresponding to different types of indefinites and definites also go on the article. 

What more can we throw onto these auxiliary nouns to get a sort of "polysynthetic noun phrase" going? Case could be an obvious thing, but let's not go there. Other kinds of relations than possession? E.g. spatial relations with regards to other nouns and deictics?

A noun that is represented by a pronominal suffix in "another noun's article" does not require an article of its own, unless it has further embedded relations (i.e. a possessor of a possessor or somesuch).

Sargaĺk: Verbs with Different Complements in the Negative

Although Sargaĺk mainly does not have differential object marking, some verbs with oblique complements have a differential thing to them.

This post lists a few of them.

ĺirjar - remember, observe any religious tradition, contemplate
     pos: loc, neg: lat
rinjas - to reach (when extending oneself towards), to stand tall, to be esteemed (worthy of)
     pos: lat, neg: abl
jaĺgan - to be welcomed into a place or situation
     pos: lat, neg: abl
garjir - to ask permission to enter into somewhere
     pos: lat, neg: abl
gardan: to grant permission to enter into somewhere
     pos: lat, neg: abl (note: recipient of permission is in the accusative)

One verb has a rather peculiar feature, where which participant is the subject changes in the negative:

pidas - fit
In the positive, a thing₁ fits into (lative) a thing₂, or a person fits (his) bodypart into (lative) a thing₂.
In the negative, a thing₂ does not fit on (locative) a thing₁, or a thing does not fit on (locative) a person on a bodypart₁

 

Detail #230: Anti-Pegative Alignment

I posted, a good while ago, a post that described the "anti-ergative" subsystem of Finnish. The post also contains an extension of this into ditransitive verbs. However, the pegative alignment that I've previously described did not enter into that post.

What would an analogous anti-pegative alignment look like, especially in a language that in the "normal" case is pegative?

Let us consider the situation when the subject is omitted:

[omitted subject] gives thing.acc person.nom

This seems like a fairly natural way of dealing with the situation - there is no case problems whatsoever. We may want to make the situation worse before we come up with something interesting. So, first we reduce the case marking a bit:

Subjects, objects and recipients are all marked by nominative, except ditransitive subjects, which are marked with the pegative.

Given the nature of, say, dechticaetiative languages, this seems like the most probable situation for a pegative alignment in a real language.

The situations we're interested in are, schematically, something like these:

1. [omitted subject] gives thing.nom person.nom
2. person.[nom? peg?] gives [omitted thing.nom] person.nom
3. person.[nom? peg?] gives thing.nom [omitted recipient]

If we extend this to a Finnic-like thing, this also would include situations where the subject is not canonically marked, although not strictly speaking omitted. We can extend this even further: non-canonical objects and non-canonical recipients.

1. Omitted or non-canonical subject
It seems this situation shouldn't call for any specific thing to happen in an anti-pegative system. Maybe, just maybe, we could have the pegative marker ascend to either the direct or the indirect object. Both possibilities seem reasonable, depending on what the speakers see as the semantics of the case marker: does it a) mark the argument with the most agency or b) the argument that provides for the recipient. To put it in very clunky English rephrasings:

a) Erin gave, and Wendy received candy.
Wendy has some agency, Erin has more agency, candy has none, thus omitting Erin, Wendy is now at the top of the agency scale.

b) From Erin, Wendy was given candy.
Wendy has agency, Erin is marked as origin. Let's use an even less natural preposition here:
Of Erin, Wendy was given candy.
Wendy was given of candy.

Both of these seem somewhat compelling (albeit probably typologically speaking very improbable). Let's present some graphs:

The standard bit of the pegative system:

SUBJ	DO	IO
nom	-------	-------
nom	nom	-------
nom	-------	nom
peg	nom	nom

Possibility 1 extends the table with these rows:

SUBJ	DO	IO
-------	nom	-------
-------	-------	nom
-------	peg	nom

 Possibility 2 does this instead:

SUBJ	DO	IO
-------	nom	-------
-------	-------	nom
-------	nom	peg

We also have possibilities 1b and 2b, where we get the following tables:

Possibility 1b:

SUBJ	DO	IO
-------	peg	-------
-------	-------	nom
-------	peg	nom

 Possibility 2b:

SUBJ	DO	IO
-------	nom	-------
-------	-------	peg
-------	nom	peg

Both of these seem pretty unlikely. If we want to see some really twisted stuff we could come up with worse varieties still, though:

SUBJ	DO	IO
nom	-------	-------
nom	nom	-------
nom	-------	nom
peg	nom	nom
-------	peg	nom
-------	-------	peg
-------	nom	-------

Or alternatively

SUBJ	DO	IO
nom	-------	-------
nom	nom	-------
nom	-------	nom
peg	nom	nom
-------	nom	peg
-------	-------	nom
-------	peg	-------

Both of these should be pretty bad options - especially due to the three last rows probably making it impossible for someone learning their first language to draw a reasonable conclusion about the distribution of the case marker, along the way that the anti-accusative system given last in my post on anti-ergative systems.

We could of course develop the details for an anti-pegative appearing in an otherwise decthicaetiative or "normal" alignment also. Such a follow-up post will probably appear in a few days.

Sargaĺk: Argument Structure-altering Constructions

Constructions that alter the number of canonical arguments include:

• the passive and the pegative-passive
• the habitual aspect (which has subject appear in the comitative, and the verb as a habitual participle)
• the construction analogous to "X had better ..."
• the distant past
• verbs of transitions

The passive in Sargalk deletes the highest-ranking nominative. There is a pegative-passive that deletes the pegative. Typically, the pegative is changed to nominative in a passive clause. This does always happen if the omitted argument instead was demoted to an oblique phrase. The passive is formed by replacing the final -VC of the citation form with -game and the pegative-passive -gagame. In the past tense, the two passives are conflated as -gven. In participles, subjects, objects and indirect objects are kept intact.

The habitual aspect usually has the participle appear clause-finally, marked by the suffix -saš (in the passive, -sanaš, in the recipient-voice, -sašen)). Thus:

Indak'-mic sunir-sa higin-saš
Indak-COM net-plur.nom tie-hab.ptcl
Indak regularly ties nets

'X had better' uses 'good' in the pegative form, sibe-ta, followed by up to three nominative arguments. Only the first of these triggers any congruence, and the verb itself takes no pegative congruence marker.

sibe-ta Indak' Mared sunir-sa tor-ju
better-from Indak Mared net-plur.nom sell-3sg
Indak had better sell Mared some nets

The distant past is formed using a few auxiliaries and the subject in the comitative.

ʒeuga (masc sg.), ʒeugi (fem sg.), ʒeugis (masc, fem pl) , ʒeugven (passive)
ʒimena (masc sg.), ʒimeni (fem sg.), ʒimenis (masc, fem pl), ʒimeven (passive)
nade (not marked for gender), nagven (passive)

Thus Jomemai Salaru seukahir ʒimena - Jome travelled to the southern lands, back in the day. Kahimai xəszin ʒeugven –Kahi was poisoned a good while ago.

The main differences are that nagven less often is used with animate subjects, and ʒimena is "further" past than ʒeuga. Ʒimena is almost exclusively used with subjects that are now dead.

The verbs of transition take the original state or the object undergoing a transition in the pegative. The most common is arda, (with the past form orga, exceptionally having a form for far past: ardana, ardani, ardanis, ardanen). Another is boyda. Boyda differs in more often implying that a previous state has been left, rather than just acquiring an additional state. It only has the past form boga) What is considered a state that has been left is rather culturally conditioned. Getting married is considered leaving the state of not being married, and thus mandates boyda. Getting sick is not considered leaving the state of being well. Getting old is considered acquiring age, not losing youth. Getting rich is considered acquiring a feature, not leaving the feature of being poor behind. Improvement in any feature mandates arda.

Jome-ta tempe orga - Jome sick became
Agu-ta pehite-rne arda - Agu chieftain-to become
Korsa-ta ganite orga - Korsa rich became
Seno-ta tusto boyda - Seno wise become
Mabi-ta ontor-rne boga - Mabi outcast became

Noun complements generally have the -rne (lative) case, adjectives are generally "naked".

Sargaĺk Non-Canonical Subjects and Objects

This post originally started out as a post describing an "anti-pegative" sub-system in a language (not necessarily a pegative language). That part of the post has been excised now, and will return in a more focused form.

In many languages, some set of verbs or some constructions cause exceptional patterns of marking. Sargalk has a handful of such verbs as well as constructions. The exceptional verbs are valjan, represent, kotjan, permit, feldar, suggest, surrender, harias, promise, pledge, grant marriage to one's daughter, merenar, replace obj1 with obj2, durenar, offer obj1 as a trade for obj2, rigmar, to lack, suldin, to assert the truth of what one's previously said, švudar, to stink, smiral, to stretch something.

In the following discussion, note that only pronouns distinguish the accusative from the nominative.

These follow a few different patterns - valjan and harian have subject = pegative, recipient = nominative, object= pegative. In constructions where any argument is omitted, the cases remain unchanged with these two verbs.

Kotjan and feldar have the direct object in the locative case, and the other two arguments are nominative. Merenar has the direct object in the pegative, and the other in the instrumental. These cases are not impervious to changes.

Harias takes the direct object in the pegative, and the other two arguments in the accusative.

Merenar has its subject in the nominative, the replaced object in the pegative, and the replacement in the lative. The subject can be omitted, and the replaced object will have verb congruence, but pegative marking. Durenar follows the same pattern, with the exception that the replacement can be in the nominative or accusative as well.

Rigmar has its subject in the pegative, and the object in the accusative. Suldin has its subject in the pegative, and no direct or indirect object appears. An eventual recipient of the assertion is in the lative, and any semantic object is really either in another VP or embedded as a subclause.

Švudar can have either the emitter of the stench or the emitted stench in the pegative, and the other NP in the accusative. Smiral has a nominative subject, pegative direct object, and nominative indirect object (generally the location to which something stretches). The (causative) subject can be omitted, turning the direct object into a new subject, which is then still marked with pegative and triggers subject congruence.

The next post will include the constructions that trigger non-canonical subject and object markings.

Detail #229: Grammaticalized "Selectedness"

A thing that could be somewhat similar to definiteness, but also to some other things nouns can distinguish. Let us consider the mental act of selection as the distinctive trait. 

Selection can be performed by the speaker, the listener, or even a third person. However, only persons aware of an object can select it, and the only relevant people for the act of selection are either speaker, listener or a third person relevant to the state of affairs described by the verb phrase.

Selection happens when the object is mentally marked out among others.

house-SEL on the hill: that house on the hill (out of many houses, or out of many potentially relevant objects)
house on the hill: a/the house on the hill

she likes snow (she generally likes snow)
she likes snow-SEL (she likes snow over the other options present)

As the house example shows, selection is not the same as definiteness: a single house on a hill is not necessarily marked as selected despite being the only thing on the hill. (However, maybe on the hill too could be marked for selectedness, thus distinguishing the full phrase house on the hill as selected, as opposed to, say, house by the beach). The snow example also shows that indefinite, mass nouns can be selected.

In imperatives, ommission of selection generally implies free choice, selection is basically the same as a demonstrative - although a fairly vague such, as some other NP involved may affect its meaning, e.g. "sell John's car-SEL" will resolve to whichever car John prefers you to sell.

Reuse of the selection morpheme refers back to the same entity as previously selected. Doubling it means a new selection takes place, after which a simple selection morpheme refers to the new selection.