Bryatesle: Adpositions vs. Location and some thoughts on Subjecthood and Adpositions

I will be talking about a technical notion of subject for a lot of this post, but at times, I will need to also refer to a less technical notion of subject. Not quite topic, since it's clearly not a role that can be filled by any old NP, nor necessary is it the topical NP. It rather is some vague notion of the quality that distinguishes the NP which in some way is active, whose state or change thereof the verb describes or whose active part in the interaction with some other NP or itself is expressed by the verb. Subjects are defined by certain subjecthood tests, but would almost all exhibit that vague notion (except maybe the subjects of passive verbs), but not all nouns that exhibit these vague notions are strict subjects, nor would pass subjecthood tests.

I will call any NP that carries these vague notions - including proper subjects - subjectids, and any subjectid that is not a subject is a subjectoid.

Proto-BDS did not strictly speaking have grammatical subjects, and their evolution in all the daughter languages showcases traces of the pre-subject state of the language. This statement requires a rather technical notion of what a subject even is, but for now, let's look at what NPs in Bryatesle with any kind of certainty are proper subjects, starting at the most certain:

1st and 2nd person pronouns that trigger verb congruence
proper nouns pertaining to humans that trigger verb congruence

These are, in all analyses, beyond doubt  as far as subjecthood goes.

Third person pronouns with human referents that trigger verb congruence are almost certain to be subjects as well

Here, congruence only really helps determine those cases where explicit congruence is present, e.g. plurals or non-neuters. (The neuter congruence is in fact a kind of zero congruence, as can be seen with the many instances where a verb gets neuter congruence despite no neuter argument being present).

A verb lacking a subject does not necessarily signify that no one or nothing is doing that, or even that who- or whatever is doing is not present in the clause. It simply means there is no noun phrase present with certain syntactical properties - it cannot control reflexives, it cannot undergo certain syntactical operations, it cannot be relativized, etc.

Now, this lax subjectness can be seen with inanimate masculine or feminine subjects:

aryk	tëk	-u	të-	-tyrn	-ai
winter	you	acc	RESLT	kill	3sg neuter
winter	you		(will)	kill

winter will kill you

The non-subjectness of aryk, winter, is not obvious from the example above, but  if we change the structure of the clause a bit we find for instance, the following permissible construction:

ark-ity tëk-u të-tyrn-ai

ark	-ity	tëk	-u	të	-tyrn	-ai
winter	ABL	you	ACC	RSLT	kill	3sg neuter
winter	-from	you		(will)	kill

winter will kill you

It turns out most inanimate nouns can be somewhat subject-like even in the ablative and dative cases.

A similar lack of subjecthood even occurs with intransitive verbs, and we find, for instance, that verbs like 'cease, end', 'begin' or 'last' basically can take an inanimate subjectoid in any case, but is more picky with the case marking on animate subjects. Plural animates may also behave as subjectoids at times, and this seems to correlate with the extent to which the plural animates act as a group or as a bunch of independent agents. Funnily enough, the 'independent agents' end of the spectrum behaves more like inanimate subjectoids than like animate subjects; it seems there is, among Bryatesle speakers a sense that the less internally organized a plural NP is, the less it is like an animate NP.

Neuter nouns fall even lower, however, on the rank of animacy, than inanimate feminine and masculine nouns, and it seems this is the reason why a separate ergative construction has emerged specifically for them. The assumption has been so strong when parsing that a neuter noun is an object if the verb is not intransitive. We shall divert our attention for a while to the formation of the ergative case:

Neuter nouns, when subjects of transitive verbs, take a masculine nominative pronoun as a particle. This pronoun is somewhat phonologically reduced, and comes immediately before the noun. Adjectives preceding the noun mark masculine congruence.

A piece of evidence that quirky case subjectoids are proper subjects in Bryatesle emerge in some dialects: quirky case neuter noun subjects of ~transitive verbs in fact also take the masculine pronoun – in some dialects in the appropriate quirky case, in some only the noun (and adjectives!) are marked for the appropriate quirky case. This could arguably be called 'quirky ergative case'.

Now that we have looked at the notions of subjectids and subjectoids, let's delve further into a different relation that oftentimes is one between two nouns; that encoded by adpositions. An adposition can relate not just a verb to a noun, e.g. telling the location of a verb's occurrence, but also of nouns related to that verb, or even more specifically, telling us about the location or direction of a noun. Like with subjectids, we sometimes get nouns displaced from their adposition, either due to them being fronted as topics, or due to some other noun being more strongly attracted to the adposition. Maybe we could name this relation anchors in lack of any better term. We get a similar set of anchors, anchoroids, anchorids. 

The oblique/obliquid/obliquoid generally sort of is analogous to the object of a verb phrase, but sometimes, an adposition also has something similar to a subject as well - e.g. in simple statements of where something is - "John is in Western Papua". In English - and mostly in Bryatesle - such NPs are not just similar to subjects, they are subjects. However, we do get situations where the notional subject of the adpositional phrase is not the subject of the VP:

I put the bottles in the refrigerator
The children found berries in the forest

Bryatesle has a tendency of not wanting to have the topic be the object of an adposition, but it also has a tendency of not wanting to have adpositions without NPs. Thus, if we were to topicalize 'refrigerator', we would get the following transformations:

I put [bottle.acc.] [[refrigerator.dat.def] in]
→ refrigerator.dat.def I [put bottle.acc] [[_____(.dat.def)] in]
→ refrigerator.dat.def I [Ø →] [[put bottle.dat] in] 

Secondary case is not as closely tied to morphosyntax as primary case is, and so does not carry over, whereas the primary case is morphosyntactical in nature, and therefore the erstwhile 'subject' of the postposition now does adopt the case the object previously had, but usually remains on the left-hand side of the adposition.

In some sense, verbs and objects behave in a similar manner here: verbs push certain types of subjectids away from being actual subjects, but rather some kind of oblique argument with subject-like properties. Somethings, postpositions push anchorids away from being anchors and into being oblique objects with anchor-like properties.

However, looking at it from a different direction they seem very different:

Verbs permit non-subject subjectoid nouns to be parsed as agents, and do not require syntactic gaps to be filled. Postpositions do not permit gaps, but permit anchors to become objects in order to fill them.

This treatment is probably a bit too technical, but this should be read as a policy statement rather than an actual grammatical treatment. This is a post clearing up some of my thoughts on this issue, attempting to form a coherent idea of the Bryatesle subjects