Friday, July 25, 2014

Detail #92: Adjective case congruence blocking

Imagine, if you will, a language where adjectives and nouns normally have case and number congruence, and adjectives precede nouns. However, genitives and some other attributes may appear anywhere  among the adjectives, thus:

John's red house
red John's house
Some difference in meaning is possible with regards to these: in a situation where a distinction between alternatives is being made, the foremost marks the distinguishing attribute. That is "John's red house" pinpoints that out of the red houses, it is John's house that is of interest. The other order pinpoints that out of John's houses, it's the red one that is being referred to.

Now, in this language, there's a further complication: for cases other than nominative, the case congruence fails whenever a genitive is inserted between an adjective and a noun, and the adjective is marked for the default case (viz. the accusative in the language I am imagining, although nominative could possibly be a good option as well - with the accusative as a default case, there's just an additional possible quirk: nominative congruence doesn't fail!)
John's red-at house-at
red-acc John's house-at
However, the genitive is also used for the first noun in apposition in compound-like lexical units composed of two nouns. In such a case, congruence is not affected.

This idea is basically inspired by various weird congruence things in Finnish:

  • an almost closed class of expressions where the adjective or determiner is in any number of cases, but the noun is in either the partitive or the instructive (~instrumental). Admittedly, this is pretty different from the idea given, but it's still an interesting non-standard congruence thing. Examples of such expressions are "pitkäksi aikaa" (in a long time, long-translative time-partitive), "monella tavoin" (in many ways, many-on manner-instructive)
  • weird things when numerals and determiners in the nominative appear: viisi miestä (five-(nom/acc) man-(part), nämä kahdet autot (these two-plur.nom/acc car.plur.nom/acc), nämä kahta autoa (these two-sg.part car.part)
The latter might be more relevant to this, but that's also as far as I can tell one of the things where native intuitions for Finnish speakers vary the most. 

Wednesday, July 16, 2014

Detail #91: Case forms referring to other nouns and derivative case

Consider the English construction "the [adjective] one". Now, English also permits a similar means of reference with genitives - "My dog is well-behaved while John's, quite frankly, is a menace"

English has a particular nice quirk with regards to this - one that I am not familiar with any other European language having, viz. possessive pronouns distinguishing when they're determiners (my car) or when they're standing alone (mine's bigger than yours). In other languages, the same distinction clearly is maintained, but not by ways that make the independent form distinct.

Let's go on and consider how other kinds of cases also can be used attributively - the bottle on the shelf, the house by the gorge, etc. So, how about permitting a kind of extended use of pretty much all the cases. This can get a bit tricky.

Suddenly, a sentence like "by the gorge burned down last night" could make sense in the right context (since everyone in the village knows about there being that house referred to as by the gorge). How do we disambiguate this? How do we know we didn't just miss the proper subject? And here we reach my usual trick - discongruence.

If we have some noun class system combined with a case system, let's just drop the class congruence. I might use this in Tatediem, but have non-nominatives trigger the 'grammatical class' on verbs. 

This also opens up the chance of using case as a kind of derivative thing - you could easily imagine lots of place-names to have derived from locative cases, house-names from genitives or locatives, comitatives being used to designate spouses (in the case that the name is not yet known or for some reason the other spouse is favored for whatever reason - sexism, obviously, could be a cultural trait that could lead to such). 

How do we deal with these when needed in other contexts than as subjects or objects with either entirely omitted congruence or some special congruence marker (along the lines of the Tatediem "grammatical class", which also contains a bunch of other things such as numbers, conjunctions-as-nouns, etc)?

If the language is mainly verb-framed, the problem should be solved quite well - if you're approaching "at the gorge", you'll probably be able to contextually figure out whether it's actually the house at the gorge or the gorge itself you are approaching - with many possible cases having both be true, with the caveat that approaching the house probably is the more salient true meaning. Obviously, you could also have the language omit the case (in favor of accusative or whatever) when you're approaching the gorge itself, and go with the locative only when using it in a locative-derivative manner. 

Satellite-framing obviously would work less well, unless you go in for the rather inelegant solution of having double case marking or adpositions that treat case-inflected nouns as though they were entirely run-off-the-mill NPs. 

Verbs of movement and such can of course have further locative things happening - you may be approaching something at the gorge. In this case, I figure a language with a system along these lines would prefer some parsing rule - either the direct object is parsed as what is being approached and the location is parsed as a locative adverb; alternatively, whichever thing is closest to the verb is assumed to be the direct object. Or finally, in the case you're approaching at the gorge at the birch forest (which would seem weird in the small imaginary village I have as a mental model right now for the speakers of this language, since the house at the gorge is nowhere near the birch forest, so the speaker must be narrating a dream or something he was confused about, or maybe at the gorge was being pulled by horses to a new place, where it'd still retain its name, now a misnomer), there might be a hierarchy of which case-inflected noun to parse as the approached object and which to parse as the wider location - or it might be affected by more pragmatics-related issues, such as whether at the gorge already has been established in the discussion as having an antecedent. This will leave gaps in what you can approach where (or whatever other verb whatever other object at/with/by/... whatever other location or thing, obviously, but that is also an interesting thing to consider more generally.)