Archive for the ‘grammar’ Category

Detail 425: Mixed groups vs. Gendered Plurals

Sunday, April 24th, 2022

In some languages, gender is distinguished in the plural. This naturally brings along the problem of what to do about mixed groups. It is fairly usual for mixed groups to default to a gender (all instances I know of default to masculine). What if a language behaved differently? What options are there? Here are a few with some hints at avenues of making the situation even more complex.

1) Free selection

The decision to use a feminine or masculine plural pronoun for mixed groups is left completely to the speaker.

2) Controlled semantic selection

Which pronoun is determined by some semantic fact about the utterance: maybe the gender indicates attitudes to the group, or maybe it correlates to TAM. Maybe actual, definite, pre-defined groups get feminine, whereas yet-undefined, hypothetical, future, potential groups get masculine.

3) Syntactic selection

E.g:

  1. Subjects - feminine, other constituents: masculine.
  2. Main clauses: feminine, subclauses: masculine.
  3. Erg-abs part of grammar: feminine, nom-acc part of grammar: masculine.
  4. Certain verbs' or adjectives have congruence that is, for morphophonological reasons, defective: singular masculine and singular plural is conflated in the verb. In such cases, the feminine is used. In some other verbs, the opposite problem applies, and so the masculine is used.

4) Registral selection

The strategy may vary by register, but it may also be as easy as in formal registers, mixed groups are masculine, in other registers, they're feminine (or maybe free or whatever).

5) Speaker- or listener-based selection

Maybe mixed gender groups always get the opposite gender to the speaker (or the same), or maybe it's the listener's gender that determines. In case of mixed listeners, ...

6) Lexically controlled selection

Maybe some verbs prefer feminine pronouns for mixed groups, some verbs prefer masculine pronouns. The deficiency in congruence in 3.4 could easily be lexicalized and stop being specific to forms of the verb where the congruence fails.

7) Referent-affected selection

The group or some individual of the group, and some property of said person(s) affects the choice: majority female gets feminine, majority masculine gets masculine, or most socially prominent member determines gender of the group.

8) Feminine- and masculine mixed groups as separate referents

I am not sure this even could evolve, but imagine a system whereby e.g. masculine plural for a mixed group essentially is proximative, and feminine essentially is an obviative pronoun.

9) No mixing

Instead of a single pronoun, two are used: "they(fem) and he", "they(fem) and they(masc)".

A Question Regarding Tone

Friday, April 22nd, 2022

Tone is normally suprasegmental. Is there any language where tone distinguishes only a few segments?

In other words, is there any language, where the vowel system has one or two vowel qualities where vowel phonemes are distinguished by tone? Something like this:



Covert Grue

Friday, April 22nd, 2022

There is extensive literature on basic color terms. Since Kílta is a personal language for speaking in the modern world, it has a fairly wide color vocabulary, and does distinguish blue and green (pikwautin, ralin), unlike a grue (green-blue) language which unifies those colors under one term.

One thing I've done in Kílta, inspired in part by the articles in The Aesthetics of Grammar: Sound and Meaning in the Languages of Mainland Southeast Asia (Jeffrey P. Williams, editor), is to pay a lot of attention to how words are intensified. English of course has plenty of intensifying collocations — hopping mad, deeply concerned, etc. — but in Kílta there are quite a few intensifiers which only intensify. They have no independent meaning, and are often (apparently) root words.

A new intensifier I recently added is . It is only used with hichínin black, pikwautin blue, and ralin green. So, even though Kílta is not a grue language, I've hidden a grue tendency in the use of this intensifier.

Ummul në mó ralin no.
forest TOP deep green be.PFV
The forest is a deep green.

Mó hichínin mika në ël si alincho.
deep black stone TOP 3SG ACC shun
The jet black stone slipped from her grasp.

I extended in one other direction. Even though it is rather adverb-like, I permit it with kinta night to mean something like in the dark of night, for in a temporal adverb sense.

Ha në mó kinta otta si cholat oto vukai.
1SG TOP deep night sound ACC hear.INF fall.PFV DISAPPR
I happened to hear a sound at darkest part of the night.

Covert boundaries can be a useful way to think new things through.

A few Terms of Time in Ćwarmin, Ŋʒädär and cognate languages

Thursday, April 21st, 2022
This is a bit preliminary.

Ćwarmin and Rasmjinj have borrowed the system of times of day, as well as higher-order calendarical structures from Bryatesle, whereas Ətimin and Astami conserve the Ćwarmin-Ŋʒädär system intact. However, the lexemes in Ćwarmin and Rasmjinj are often cognate to those of the other Ćwarminoid languages, and have simply been repurposed to Bryatesle cultural standards.

Historical Background
The Ćwarmin-Ŋʒädär ur-tribes divided the day into a system somewhat familiar to us: morning, day, evening and night. The 'cycle' is considered to begin at sunrise. Every ĆŊD language has cognates to at least some of these terms. The divisions correspond to watches in camps (which, however, normally overlap.)

Subdivisions of the day
An important pair of adverbs that have surviving cognates in every branch, and also 'functional' equivalents in many languages that lack cognates, are
*birn
*voru
which signify 'in the previous quarter of the day' or 'in the next quarter of the day'. The proto-language formed words meaning "a day and a quarter from now" or "three quarters of a day from now" by reduplicating these. Sound-changes have hit these particular words in some special ways - sometimes, the morphemes have been treated as separate words until at some point in each language's history, they've become full-fledged single morpheme words.
Cw: birmir, uarjur ( b > v / r_V or #_u,  v > j /r_V,  > u #_V)
Ast: birbir, uoruor (v > u /_o and also v > u /r_)
Rs: birnə, vorbʊ (v > b /_uC, _oC, _ʊC)
Ŋʒ: vırmız, varoz ( b > v /#_V, -z is a suffix)
Dg: (m)ʊmber, (m)ʊrbel (#b > #m, randomly. rv > rb, murber > murbel due to dissimilation)

In the Ćwarmin branch, a similar time yesterday can be specified by suffixing *-zi or *-zu. This -zi suffix is probably cognate to Ŋʒ -z. in Ćwarmin and Rasmjinj these terms exist, but now refer to the next span of Bryatesle day subdivisions.

A verb 'birdən' signifying 'to wait for one's turn, to be preparing for a task, to expect, to soon be busy' derives from *birn, whereas 

In Ćwarmin and Rasmjinj these words rather signify the next/previous Bryatesle unit of time, of which there are eight per day, determined by the inner moon and the sun.
 

Morning, Day, Evening, Night

Ćw Ast Rs Ŋʒ Dg
arad arro arot äzä adʒı
jit int injtjin uk'u imii ımı/(f)ındı-)

The proto-ĆŊD word for morning was *azda, potentially either cognate to *asu (to wake up) or *anzor (sunrise), probably related through some earlier word, maybe pre-proto-CND *egnso- 'rise up, raise, open (of eyes, bottles, jars), burst'
Ćw:  arad (azda > azad, z > r /V_V)
Ast: orro (azda > aza > a'ra > awr:a)
Rs: arot (azad > azod > azot > arot)
Ŋʒ: äzä (azda > aza > random vowel harmony realigment)
Dg: adʒı (azda > adza,  a > ı / _#, dz > dʒ / _ı, _i) (change in meaning: 'early')
Proto-ĆŊD for 'day' was '*gnumn', giving
Ćw: now (um > om, m > w /V_n#, where V = back vowel)
Ast: navvol
Rs: naon
Ŋʒ: ŋumor ('tomorrow')
Dg: ŋö(-n-)

*gnumn may be cognate to pre-proto-CND *gunu- sky and thus be related to Dg ŋost ('cloud'), Ŋʒ ŋuzro ('arctic lights'), Ast narrob ('rain'), Cwarmin owno (sky), Ast owno

Some derived terms:

Astimin and Rasjminj also derive recent words for the sun from this:

ast: navvark, rs: naoroh

A similar word can be found in some poetic Ćwarmin: noworak

Proto-ĆŊD for evening was *tyrs, giving
Ćw: tic (y > i / _r, rs > c /_#, )Ast: ter (i > e /#(C)_r#)
Rs: tils (rs > ls, random, y > i, random (~0.4 prob) in monosyllables)
Ŋʒ: tydźy   (rs > rź, rź > dź, -y is a nominalizing suffix)
Dg: jyrem (t > j / #_Vr, V = front rounded vowel, probs by proxy of t_j_w > d_j_w > d_j > dʒ, -em = time affix)


The proto-ĆŊD word for night was *imid. In parts of the Ŋʒädär branch this has been replaced by reflexes of *uk'ot, 'dark'. Thus, 'night', sometimes with varieties such as 'this (incoming) night' or 'last night':
Ćw: jit (but jint- before suffixes that begin with vowels)
Ast: int ('inni' for 'this (incoming night', 'inits' for 'last night')
Rs: injtjen
Ŋʒ: uk'u (no ending in absolutive, -s-/-t- in other cases).
"imi-" now signifies sunset instead.

Süw: imii
Dg: ımə, dat. fındın or ındın. (The case prefix has acquired some meaning differentiation in expressions of time and been generalized to all cases except the absolutive, and distinguishes "night (in general)" from any particular night.

Today, Yesterday, Tomorrow and Beyond
In all branches there are languages that preserve cognates of a variety of old words.

day before yesterday: *qaluwuna
yesterday: *qalur
today: *mest
tomorrow: *tetri
day after tomorrow: *tetrijinä

Astami: N/A, kalu-nu, mih-ni, ćeič-ni, ćeiči-ni
Dagurib: qaluwu, qalu, mesit, tetir, tetirjin
Ŋʒädär: qoruŋa, qorur, mär* (mäsä*- for inflected forms), täryr, tärynjä
Süw: qurquur (reduplicated), quur, met, teir, tädeir (through intermediate forms *terteer > *terdeer > *terdeir)

*mär, mäsä have been lost in modern Ŋʒädär, being replaced with ŋumrum, from ŋumor (to-day). Almost all speakers tend to dissimilate either the last -m to n, giving ŋumrun, or m to b, giving ŋubrum or even ŋ to g, giving gumrum. This serves to distinguish "to (a/the) day (indefinite)" from "today" - "to a day" being regularly formed and "today" having those sound changes.

However, in early Ŋʒädär, *mär, mäsä still were present. Tärynjä no longer is specifically the day after tomorrow, but any day in the near future (including, possibly, tomorrow).

Süw's tädeir likewise does not signify 'the day after tomorrow', but is an adjective signifying 'the next [timespan]'.
 
 In Ćwarmin, 'today' is formed form the demonstrative arna- in the general ablative: arnaraś. Sometimes this is hit by some kind of dissimilation-transposition and comes out as aranaś, anaraś or even arnaś but there's also attestations of both araraś and ananaś. A period of a few days including today can be arnuroś / arunoś. Oftentimes, this signifies 'this week' (by Bryatesle standards of week). In Bryatesle-influenced areas, you also often get olbaraś (from olba, "that") for 'yesterday'. This is somewhat odd, though, as it can also signify 'that day' or even 'that time', and so is a bit sensitive to context.

Despite being marked for case, aranaś and arunoś have partially been reinterpreted as nominatives, and can take further case suffixes.

'In a few days' in Ćwarmin is generally formed by cularaś (sg) or culuroś (pl), depending on whether the thing that happens is expected to last at most a day, or longer. Similar constructions exist in the other Ćwarminoid languages.


Verum focus

Saturday, April 9th, 2022
With the ordinary kind of focus that we've been talking about all these years, we're identifying a constituent that's new or important to the discourse. It's also important, though, to be able to focalize the truth value of an utterance. I have notes in my Koa journal (mainly vaguely worried questions) about this concept going back several years, but only recently started thinking about it an organized way.

Though Describing Morphosyntax termed this "truth value focus," quite a lot of research last month informed me that the best technical term these days is "verum focus," or just "verum" as some people are very passionately willing to argue. I was gearing up for some major construction when I realized that Koa actually already has a built-in way to do this! Let's first look at a pragmatically neutral clause in AFF/NEG/INT forms:

ni te puhu le níkili
1SG ABIL speak NAME English
"I speak English"

ni na te puhu le níkili
1SG NEG ABIL speak NAME English
"I don't speak English"

ai se te puhu le níkili?
QU 2SG ABIL speak NAME English
"Do you speak English?"

The simplest way of focusing on verum is via the particle ia, initially conceived as a firsthand experience or vouched-for evidential but now clearly functioning as as a viridical marker. It shifts the primary purpose of the utterance from the semantics of the constituents to a confirmation by the speaker of the utterance's truth value. As such one would expect that the clause to which it's attached would not contain any new information, since the focus, so to speak, is on verum in the context of a discourse stage with existing players; it would be anomalous if used without that existing context, or would at least cause the listener to infer that there was some existing context of which they were unaware. With our sample clauses from above, then:

ni ia te puhu le níkili
1SG VIR ABIL speak NAME English
"I DO speak English"

ni ia na te puhu le níkili
1SG VIR NEG ABIL speak NAME English
"I DON'T speak English"

ai se ia te puhu le níkili?
QU 2SG VIR ABIL speak NAME English
"DO you speak English?"

A note on accentuation: in AFF and INT contexts the main stress is on the ia above: ni iá te puhu... In NEG contexts, though, the stress in ia na is on na, and in fact they may be written together and accented to make this plain: ni ianá te puhu...

We can also get at this concept periphrastically with eso "real, actual, so" and a dependent clause:

eso ko ni te puhu le níkili
real COMP 1SG ABIL speak NAME English
"it is the case that I speak English"

na eso ko ni te puhu le níkili
NEG real COMP 1SG ABIL speak NAME English
"it is not the case that I speak English"

ai eso ko se te puhu le níkili?
QU real COMP 2SG ABIL speak NAME English
"is it the case that you speak English?"

These are pragmatically neutral again, though, without any particular focus. We can ratchet things up or add focus in a few different ways depending on how heavy-handed we want to get (translations here are kind of stilted -- real idiomatic English would of course use a variety of words and also intonation to get at the meaning: "no, look, I told you, I DO speak English," etc.):

eso sa ko ni te puhu le níkili
real FOC COMP 1SG ABIL speak NAME English
"the thing that's the case is that I speak English"

ia eso ko ni te puhu le níkili
VIR real COMP 1SG ABIL speak NAME English
"it IS the case that I speak English"

ia eso sa ko ni te puhu le níkili
VIR real FOC COMP 1SG ABIL speak NAME English
"the thing that IS the case is that I speak English"

Eso can also be used serially to mean "really/actually X," which introduces an interesting distinction we can make here.

ai se ia loha ni?
QU SG VIR love 1SG
"DO you love me?"

ai se loha ni i eso?
QU 2SG love 1SG VP real
"do you really love me?"

The translations might communicate what's going on to a native English speaker, but context is really critical in explaining the difference. In the first question with ai, the speaker has significant doubt as to whether the proposition is true, probably even predicting a negative answer. There's a sense of "tell me the truth, I need to know, I can take it." In the second sentence with eso, the speaker either thinks or hopes that the proposition is or may be true, and is seeking confirmation or reassurance.

It's interesting that the syntactic structure for verum focus is entirely different from that of constituent focus, but I think that's okay given that Gutzmann et al. claim that verum focus isn't really focus anyway; it seems like many languages have different structures for these kinds of emphasis.

Detail #424: Gratuitous Use of Reduplication

Friday, April 8th, 2022

One morphological device that I keep wanting to use, but never find a sufficiently interesting use for, is reduplication. Let's try and find a really gratuitous use of it, and overviewing some of the strange things languages do with it.

Some of the trivial stuff reduplication does is:

  • form plurals
  • form habituals, form perfectives
  • form intensives, diminutives, etc

The strangest use I have come across is Chukchi: the absolutive singular for some nouns is formed by reduplication. This violates two proposed universals, so that's a lot of bang for a buck!

So, what other weird thing could we use reduplication for? It feels like this is a question where the usual suspects don't quite cut it.

Let's assume, unless otherwise specified, that I am talking about full reduplication of a lexeme.

1) Things with numerals (numeral symbols express the actual value, letters express the way it's said in the language, base ten is assumed but this is a trivial thing to reapply to some other base.)

one = 1
oneone = 11
two = 2
twotwo = 12
three = 3
threethree = 13

...

oneoneone = 21
twotwotwo = 22
threethreethree = 23

With reasonably short numerals, this isn't even particularly clumsy. Heck, you can have some pretty big numbers before running into finnish-style numeral length (kaksikymmentäkaksi = twotwotwo).

With just a few extra tricks - say, having a dedicated short form for some particular milestones, this wouldn't be unworkable. 

2) Indefiniteness

Have reduplicated nouns signify "any old ...".

(Somehow, it seems this would be rather natural with some type of intonation pattern).

3) Reflexivity by reduplication of the verb

I see see in a mirror = I see myself in a mirror

4) Reflexivity of possession by reduplicating the object:

I met wife wife when I was twenty three
I met my wife when I was twenty three

he called brother brother
he called his brother

5) Comparatives

Double the comparand which is characterized by more of the quality that is compared, use some special conjunction or just apposition for the comparands or maybe object marking or something:

I I he are strong: I am stronger than he

I I am strong him: I am stronger than he

For "oblique comparisons", try this on for size:

I I am smart smart him strong: I am smarter than he is strong.

6) Extend the reference of the subject (or maybe some other constituent) by doubling the verb:
I eat eat: me and my associates are eating

I approach approach house: I and my associates are approaching a house / I am approaching a village

7) Ordinals

man man = the first man
man man man = the second man

I imagine this could actually exist for a few lexemes in some actual language!

8) Copula! E.g.

it red red: it is red

This also leads to an interesting thing w.r.t. verbs - maybe 

it eat eat = it is edible

9) Non-referentiality!

So, one example of a non-referential pronoun is "it" in "it is raining". Imagine a language where this has to be "it it is raining".

10) Adjectives denoting being in possession of something, e.g.

peg-leg peg-leg man: peg-legged man
or maybe just
peg-leg-leg man

In a language where it's done by just reduplicating a syllable, this does not seem particularly out of the ordinary.

11) Mandatory reduplication of initial and final elements of parenthetical statements as a form of bracketing.

12) Particles of phrase verbs (either the verb or the particle needs doubling)

13) Vocatives

This seems a case that reasonably could have developed a reduplicated form in some language in the world.

14) Wherever the syntax has a null element that is actually syntactically present (e.g. omitted subordinating conjunction), a floating reduplication emerges that needs to find a host:

I didn't know that she's famous -> I didn't know __ she's famous ->
"I didn't know know she's famous" or "I didn't know she's she's famous".

Under some circumstances, other syntactic phenomena could shuffle where this turns up in unexpected ways.

15) Congruence with a certain noun class by reduplicating verbs or adjectives or pronouns.


Undoubtedly, stranger ideas are possible.

Topics in Koa as a Second Language

Monday, April 4th, 2022

My seven-year-old daughter Eleanor seems to be a born linguist -- a phonetician in particular -- and it's been fascinating and heartwarming in a way I definitely never expected to experience as she's gotten increasingly interested in Koa and started making an effort to speak it with me over the past six months.

For some reason bedtime tends to be when Koa drifts into our conversations, and it's impossible to describe the emotions when this little daughter of mine says unprompted -- as she often does now -- Ivo koa, mama. Ni loha se! "Goodnight, mama. I love you!" Recently she started counting our goodnight kisses and hugs: ena, lua, tatu, nei...

Today Eleanor took in the fact that Koa adjectives come after their nouns instead of before, and after a moment of amused delight, seemingly accepted and integrated it as she started to use that syntax effortlessly (anu kuma "hot water"). Last night she struggled for a moment with Koa's distinction between singular and plural "you," but again mastered it seemingly without effort. I've never been in a language class with kids this young before, and though with a lifetime of linguistics background I know about ease of learning at a younger age in theory, I've never seen this myself and it's just astonishing.

What's most salient to me, though, is what I'm learning about my own language's phonology through my daughter's ear. Eleanor has always been unusually perceptive about phonetics both in hearing and production (I remember her at 3 explaining the difference between Spanish and English no in a way that would have fit right in in an undergraduate linguistics class) and in her Koa I've discovered the following so far:

* Koa stops are unaspirated
* /o/ is closer than I might have expected, as she's occasionally confusing it with /u/
* My /l/ apparently sometimes approaches some kind of alveolar flap! This isn't necessarily surprising in theory given that /l/ is Koa's only liquid, but I certainly never did this intentionally.

I'm really hoping Eleanor might be able to help me figure out what's going on with /h/, actually, which I've been noticing some instability with lately. Depending on the day, it seems like one of three scenarios is happening: (1) that the onset of voicing in a /hV/ sequence is delayed potentially throughout the duration of the vowel, to the point that hake "search" can come out sounding like [ḁke] or even [ḁʰke]; or (2) that the onset of voicing occurs early, voicing the onset as in [ɦake]; or (3) sort of a combination of the two, where the /h/ is colored into breathy voice which then spreads through the vowel à la Gujarati: [a̤ke]. I'm very curious to discover what's apparently already occurred in the "natural" development of this language!

Possibly even more thrilling, Eleanor -- who already loves to "invent" fairy and animal languages, speaking them in fluent and phonologically coherent gibberish -- helped me create her first word today! She wanted to say "safety" but I didn't have a "safe" root yet, so she suggested lapa and I entered it into the books then and there. She continues to seem genuinely excited about the opportunity to contribute to the creation of this language, and I'm equally excited to suddenly have such an enthusiastic partner in what has until now been an entirely solo pursuit. I just could never have imagined that my daughter and I might find ourselves here.

Incidentally, "Koa as a Second Language" would seem to translate as le Koa mo Cimi Lúasi (KCL), should there ever be a need to refer to this important subject more efficiently.

Designing an Artificial Language: Metaphor

Friday, April 1st, 2022

Rick Morneau is a long-time language creator who lives in rural Idaho. In the early 1990s, he wrote a series of essays on language design that proved to be quite influential in the early language creation community. Their quality has endured since their original publication, and continue to be read and enjoyed by language creators the world over.

Abstract

This essay is about the use of metaphor in ALs, why they should be avoided, and how they can be avoided.

Version History


This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.

Presentation of the Duišan and Kven language

Tuesday, March 1st, 2022

Sébastien Thomine just completed a master in Indigenous Studies at the university of Tromsø in Northern Norway. Currently enrolled in a second year bachelor in Kven, a minority language from Northern Norway, he is also a PhD applicant in the field of cultural representation in education and has a great interest in the revitalization of endangered languages.

Abstract

This paper is a revised version of the final exam of the course Constructed Language of the University of Tromsø. It compares the particularities of the constructed language Duišan with the particularities of Kven, a minority language derived from Finnish and spoken by less than 2,000 individuals in Northern Norway.

Version History

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License

Real Language Examples: Incongruent Expressions in Finnish

Sunday, February 27th, 2022

I recently read a PhD dissertation from the late 80s about the development of incongruent expressions in Finnish and other Baltic-Finnic languages, as well as Sami. This may well be an interesting topic for my readers, and since the dissertation is not available for sale anywhere, I figure I may as well present a summary of it.

Baltic-Finnic has adjective congruence, with the same morphemes on both adjectives and nouns. Here are some examples from Finnish:

uude-ssa talo-ssa
vanha-lla tori-lla
punaise-t auto-t
vanho-i-sta kirjo-i-sta

The other languages are similar, with some exceptions for really recent cases, e.g. in Estonian. In Estonian, recent cases derive rather naturally from postpositions, and the postposition has not (yet?) spread to the adjective. AFAICT, the adjective is in the case that the postposition previously ~governed, but I may be wrong on this.

In at least some Sami languages, demonstratives and interrogative pronouns also have some amount of case congruence, but the adjectives in Sami in general behave a bit differently, with attributive and non-attributive forms.

The incongruent expressions are a semi-productive set of expressions in Baltic-Finnic where the case congruence is mismatched. Not only that, sometimes the number congruence is off. For the number congruence, it is relevant to know that the instructive is often a case with some amount of defectiveness: most nouns lack a singular instructive, and arguably it's borderline an adverb derivation rather than a case. It is also historically probably the same case as the genitive, with some intriguing complications along that line.

Nearly all of these have either the instructive or partitive as the case of the head noun. The adjective or the determiner is usually in some local case, such as the abessive, allative, elative, ... 

Examples:

pitkä-ksi aika-a
mui-na aiko-i-n
näi-ssä ma-in
tuo-lla pä-i-n
näi-ssä määr-i-n
tuo-lla tapa-a
tuo-lla tavo-i-n

The question that the dissertation attempts to answer is how such a situation has come about. It does this by also investigating the situation and statistical situation of the expressions in several closely related languages.

It turns out that the expressions probably can be split up into several subtypes based on their semantics. Time, location, amount, manner, [[state or position] of [body-parts, clothing or mind]].

A closely related question that intertwines with this is how did congruence emerge?

Most authorities on the topic seem to agree that Indo-European has been an important influence in its emergence, but also that apposition has been a factor. Imagine that sometimes, for emphasis "catch a big fish" has been expressed "catch a big (one), a fish".

Adjectives being used as nouns by utilizing case suffixes is well established in congruence-less branches of Uralic, and in other families of agglutinating languages that utilize cases and have adjectives.

This apposition for emphasis - "in a small one, in a cave" may well, as time has passed on and indo-europeans in the vicinity have had a similar phenomenon going, have gained ground as the way to express in a small cave.

 This explains some similar expressions.

tuo-lla tapa-a
tuo-lla tavo-in

may both have appeared as a result of apposition, where there has been some level of synonymity between the two parts, but together, they have resolved some ambiguity:

tuolla signifies "with that, utilizing that, over there, on that"
tapaa, tavoin signifies "by a/the method"

This "utilizing that, by the method" > "utilizing that method".

These account for a fairly small number of expressions.

A separate type of construction that may be relevant also is called nominativus absolutus.

minä juoksin kädet ilmassa

I ran hands in-air

odotin sateessa, takki märässä
I waited in-rain, jacket in-wet
I waited in the rain, with my jacket in wet
I waited in the rain with my jacket wet

In English, this would come out as "I ran with (my) hands in the air". Apparently, earlier in Baltic Finnic, this was often more similar to English - with an oblique case instead of a pure nominative. Both instructive and partitive seems to have occurred, but unlike English and the Finnish nominativus absolutus the status or position was positioned before the noun:

ilmassa käsin/käsiä
märässä takkia/takkia
lumessa päin

Now, many of the nouns that stand on the left side in these constructions are indistinguishable from adjective forms. Märkä both means 'wetness' (noun) and 'wet' (adjective). Thus, these were sometimes reinterpreted as adjectives before nouns with a case discongruence, rather than a noun in a case acting as an attribute of another noun in another case.

This does not catch all the interesting stuff I came across in the book, and there will probably be a second post based off of it.

Source: Juha Leskinen, "Suomen kielen inkongruentit rakenteet ja niiden tausta (incongruent adjective constructions in Finnish)", 1990. Available in Finland from Varastokirjasto, and thus probably can be ordered from any library. Some university libraries undoubtedly have a copy. Availability elsewhere probably somewhat correlated with departments of Uralic linguistics. If you're really lucky maybe someone sells it on ebay or amazon or somewhere.