Scratchpad (non-Sego)

[bradrn]

[…] this leaves a role for grammaticalised reduced nouns, which now can modify other nouns to provide positional info (common adjectives may also frequently appear in reduced form).

More commonly known as ‘relational nouns’. As in your list, they often derive from body parts or geographical terms. (I have one or two survey papers somewhere if you want.)

Thank you, I’ll say yes to any papers you want to link.

OK, will send them tomorrow morning. (I may need a reminder!)

Now that I think of it again, there is a lot of interesting stuff in these papers about how spatial orientation is represented in languages. (Which is why I saved them in the first place!) One thing I remember is that some languages classify all nouns into sets, with each set requiring a different classificatory locative verb — this is found in such exotic languages as Yélî Dnye and Dutch. (Though not German or English, curiously enough, except as an optional stylistic choice: ‘the building sits at the corner of Whatchamacallit and Thingumabob Streets’.) I’m sure there’s some interesting choices for ambiguity which can result from such things. But I’d better stop typing now, because otherwise I’ll get fascinated by the subject again and will never get to sleep…

How to resolve a string like garah put hidiw ‒ antelope in front of person, or person in front of antelope?

Usually languages use possessive constructions, so it’s not an issue — as indeed is shown in your English translation: ‘in front of (the) person’.

I may be misunderstanding you here, or vice versa. To restate my question: in the context of no fixed noun phrase order, modifier-modified order, and the absence of a requirement to mark case (including genitive), how is one supposed to understand the relationship between two nouns separated by an adposition? It might be a postposition on the first or a preposition on the second. (Sorry if that is how you took my question, and it is me not receiving your answer properly.)

Oh, well… in that case, they’d be ambiguous, I guess. But isn’t that the whole point?

Since I’m trying to avoid setting one strategy for anything, I mentioned two possible strategies ‒ and you suggest one I left out, using a marked possessive construction, e.g. garahar put hidiw (antelope-GEN in.front person) ‘a person in front of an antelope’. Is that what you meant?

Yes, sort of. In a lot of languages, the predicative version translates literally to something like, ‘the person is at [the front of the antelope]’, where the relational noun ‘front’ indicates a whole region of space rather than an actual part of the antelope. Similarly you can get things like ‘the plane is at [the top of the mountain]’, where we’d say ‘the plane is above the mountain’.

For the attributive version, I’m not so sure, but I suspect it would be language-dependent. For instance, I seem to recall that French forbids a prepositional phrase from directly modifying a noun, so you have to use a relative clause — la personne qui est devant l’antilope, (though I could be wrong). I don’t know what would be best for this language.

[sasasha]

One thing I remember is that some languages classify all nouns into sets, with each set requiring a different classificatory locative verb — this is found in such exotic languages as Yélî Dnye and Dutch. (Though not German or English, curiously enough, except as an optional stylistic choice: ‘the building sits at the corner of Whatchamacallit and Thingumabob Streets’.) I’m sure there’s some interesting choices for ambiguity which can result from such things.

This is nice. I can’t see this language classifying nouns this way, but I can certainly see it permitting the use of zany locative verbs.

domoh - building, structure, home, house
rijik - to accumulate, to build, to be built, to rise
lisir - river
sislisir - stream

arijik domohi casset sislisir - our house is (lit. rises/is built) by the stream over there.

Curious to see your papers

For the attributive version, I’m not so sure, but I suspect it would be language-dependent. For instance, I seem to recall that French forbids a prepositional phrase from directly modifying a noun, so you have to use a relative clause — la personne qui est devant l’antilope, (though I could be wrong). I don’t know what would be best for this language.

It’s probably the same story of many options, few rules.

[bradrn]

Curious to see your papers

OK, here we go…

Firstly, Ameka & Levinson’s The typology and semantics of locative predicates: posturals, positionals, and other beasts. (This is actually the Introduction to a Special issue of Linguistics, all of which is extremely interesting.) I particularly like the typology presented on pp863–864 as an overview of the possibilities.

Then there’s ed. Levinson & Wilkins’s Grammars of Space: Explorations in Cognitive Diversity. This is a rather long book but with lots of detail on individual languages — each chapter is very interesting. (This is, IIRC, the book which first developed the typology discussed in the previous reference.)

On relational nouns specifically, there’s Bowden’s Behind the preposition: Grammaticalisation of locatives in Oceanic languages. I particularly like the summary tables on p11 (for African languages) and pp35–39 (for Oceanic languages). Sadly there’s not many examples of how relational nouns are actually used, though there’s some near the end (starting p59). That said, Levinson & Wilkins’s book discusses them also, most prominently in the chapters on Tzeltal, Tamil and Kilivila.

[/nɒtɛndəduːd/]

I swear, I look away for...

well I suppose it's been two days, but still! this concept of kitchen sinking (kitchen sinking??) a language with too many options and not enough rules sounds like the best kind of headache. I would love to see more of this conlang.

[sasasha]

I swear, I look away for...

well I suppose it's been two days, but still! this concept of kitchen sinking (kitchen sinking??) a language with too many options and not enough rules sounds like the best kind of headache. I would love to see more of this conlang.

Thank you for the positive feedback!

[Papers]

Thank you so much, I’m working my way through, very interesting indeed.

[sasasha]

Relative clauses

How do you purposefully kitchen sink relative clauses?

Some ideas:

1. Normalise not bothering with them, using verbal operators as serial verbs, and dealing with resultant ambiguitues.

limix - v. hide

ak garah xo set domoh noroh limix.
3.NOM antelope PAST near house run hide
An antelope ran and hid near the house.
An antelope was near the house, then ran and hid.
The antelope that was near the house ran and hid.
The antelope that was running near the house hid.

Some of these readings may be more likely than others, and non-prescribed pragmatic features may be suggestive of some over others (i.e. choice of morphosyntax might serve to topicalise elements).

2. Provide a relative marker which can be applied to the person affixes, or, linked with -a- (the 3rd person affix), to nominals. If I follow the formula of using a single consonant, I’m running low on consonants. q is a possibility.

domoh set qaa garah xo norohaan limix.
house near REL-AN-3 antelope PAST run-AN-3-COL hide
The antelope near the house ran and hid.
The antelope that is by the house got here because it ran and hid.

(I threw in some animacy marking there, and note that I’m varying how morphemes are arranged from example to example. It would be silly to convey all possible pragmatic readings of an utterance in glosses, so these are just suggestive of what the syntax, in turn, may be suggesting.)

There are no doubt other options, and other types of relative clauses. TBC

[bradrn]

1. Normalise not bothering with them, using verbal operators as serial verbs, and dealing with resultant ambiguitues.

I have Very Much to say about serial verb constructions, which I’ll (with difficulty) restrain myself from doing. Suffice it to mention that I’ve never seen one translated with a relative clause in English (that I recall). But that’s not a problem, because as far as I can tell set isn’t a verb. (On the other hand, domoh limix forms a perfectly fine SVC, which I believe would be of the cross-linguistically common ‘resultative’ variety.)

[sasasha]

1. Normalise not bothering with them, using verbal operators as serial verbs, and dealing with resultant ambiguitues.

I have Very Much to say about serial verb constructions, which I’ll (with difficulty) restrain myself from doing. Suffice it to mention that I’ve never seen one translated with a relative clause in English (that I recall). But that’s not a problem, because as far as I can tell set isn’t a verb. (On the other hand, domoh limix forms a perfectly fine SVC, which I believe would be of the cross-linguistically common ‘resultative’ variety.)

Yeess, kind of. I don’t know what is to be gained by going round in circles with the word classes conversation here as well as in several other threads. But set is not definitely not a verb. It’s reduced, it’s bleached, it can function as an adposition either predicatively or attributively, and it can be combined with the copula ‒ but I’m intending that any verb can appear with the copula (perhaps to assert positive polarity after a negative question; other uses suggest themselves to me as well). It’s arguably some kind a stative verb.

I have seen some of your work on SVCs and will probably refer to it further down the line if I keep trying to build this up.

Ultimately those weren’t the best sentences to test with.

bulum - v. eat
qadah - v. buy
kesew - n. fruit
himih - n. sibling

qadah i kesew, himih bulum a.
buy 1 fruit, sibling eat 3
My siblings ate the fruit that I bought.

buluman himihni kesewat qat qadahixo.
eat-3-COL sibling-COL-1 fruit-3-ACC REL-3-ACC buy-1-PAST
My siblings ate the fruit that I bought.

himih bulum limix a.
sibling eat hide 3
The siblings who had eaten it hid.

xo limix himihak qa ta bulum.
PAST hide sibling-3 REL-3 ACC-3 eat
The siblings who had eaten it hid.

[sasasha]

Syntax markers

Some rather strange lexemes, which may be joined later by a few others, to specify the unmarked word order of the following passage:

katac - prefixes predominate, SOV order
takac - prefixes predominate, OSV order

kacat - subject prefixes and object suffixes, SVO order
tacak - object prefixes and subject suffixes, OVS order

cakat - suffixes predominate, VSO order
catak - suffixes predominate, VOS order

-(a)r(a)- may be affixed, relative to the -(a)k(a)- morpheme, to show the general position of a modifier (-r-) to the modified (-k-).

Whether or not these are useful remains to be seen, I may scrap them, but they’re posited for now. ‘Prefixes on verbs’ btw includes morphemes separated, by choice, from the root.

Whether and how these interact with the -c- morpheme I have already mentioned is TBC.

[quinterbeck]

Taking this further, a carefully-designed combination of — let’s say — switch-reference marking + noun classes + obviation + case concord could completely eliminate the need for explicit NPs after the first mention.

At the risk of diverting the conversation away from Guqpik (which I like so far!) - why do you include case concord here? How would that aid NP dropping?

(Case concord as I understand it is when both the head and the dependents of an NP are case-marked, so do clarify if I'm off target there)

[bradrn]

Taking this further, a carefully-designed combination of — let’s say — switch-reference marking + noun classes + obviation + case concord could completely eliminate the need for explicit NPs after the first mention.

At the risk of diverting the conversation away from Guqpik (which I like so far!) - why do you include case concord here? How would that aid NP dropping?

(Case concord as I understand it is when both the head and the dependents of an NP are case-marked, so do clarify if I'm off target there)

Because it allows the relationship between the modifier and the verb to be marked explicitly, which in turn facilitates use of the modifier without its head noun. It’s definitely the ‘odd one out’ in that list, though.

[Travis B.]

Syntax markers

Some rather strange lexemes, which may be joined later by a few others, to specify the unmarked word order of the following passage:

katac - prefixes predominate, SOV order
takac - prefixes predominate, OSV order

kacat - subject prefixes and object suffixes, SVO order
tacak - object prefixes and subject suffixes, OVS order

cakat - suffixes predominate, VSO order
catak - suffixes predominate, VOS order

-(a)r(a)- may be affixed, relative to the -(a)k(a)- morpheme, to show the general position of a modifier (-r-) to the modified (-k-).

Whether or not these are useful remains to be seen, I may scrap them, but they’re posited for now. ‘Prefixes on verbs’ btw includes morphemes separated, by choice, from the root.

Whether and how these interact with the -c- morpheme I have already mentioned is TBC.

One thing to consider as an option is inverse marking without explicit marking of word order or cases, where one can have a personhood/animacy/topicality hierarchy and there is marking as to whether the object has higher personhood/animacy/topicality than the subject (with the unmarked situation being that the subject has higher personhood/animacy/topicality than the object). Combined with word order being used to indicate topicality when disambiguation is needed this can allow a much freer word order overall. I am a fan of this personally, I should note.