RFC 0003: regarding Cr

[mklcp]

I agree with the heart of the proposal and how you plan to implement it for Cr roots.
For Cs roots, I make a separate post

18-cell grid with a 24-cell grid.

Technically not exact, because for many TNIL roots, only Stem 1 has all of its Specifications explicited. So this is just a 4 Specifications + 2 other Stems + three FML = 9 cell thing. Not really a grid, since you're expected to devise the Specifications by yourself, which might be even worse, as:

• there is no rule for doing so
• if the definitions of the 4 Specifications count as "rules", then why not use those rules for Stem 1 also (if they are productive enough..)

We can extract several law-like observations from this table:

I would change the wording to "This root exemplifies several observations that can be worded as"

Further remarks

There is worse. Specification does not play well with Function.
Take for example the BSC of -G-: (to be) an instance of bodily ambulation; to ambulate

Because there is a (to be) before an instance of bodily ambulation, and because to ambulate has clearly a more verbal reading than an instance of bodily ambulation, it can be deduced that:

• an instance of bodily ambulation is the STA nominal reading
• to be an instance of bodily ambulation is the STA verbal reading
• to ambulate is the DYN reading

This is problematic in many ways:

• the nominal/verbal distinction is left undefined for the DYN reading
• for many roots, the STA nominal reading is just "a state of ...", this is not very useful, and can be easily derived via a VxCs
• for roots that have a "to be a state of ..." CTE, this is even harmful, because this prevents from transparently using any other cases than THM. The only transrelative case that is meaningful and useful with "to be a state of ..." is THM. Other cases could be used, but that would not result in something useful. For example, using ERG with a "to be a state of ..." verb would mean "(it) causes sth to be a state of ...".
  This is meaningful, but far from useful. It is way more interesting to have instead: "it causes sth to be in a state of ..."

So the default meaning, for a STA reading, should be "to be in a state of ...".

[[note: the Format solution allows that by using STA + an AFF Format]]

Having "to be a state of" is harmful because it adds a layer between the base meaning of the root, and the argument structure that Cases can access. The core slots/roles of the base root are not usable anymore, only remains the mostly ininteresting "THM is a state of ..."

hereafter called Format

Sounds a good name to me, but we should be wary of two things:

• the terminology "Format" already exists, and have another meaning, albeit a similar one
• it's better to not mess up with JQ's terminology, which will directly cause incompatibilities. If we invent our own, it might suffer from incompatibilities if JQ decides to use it for his own dark designs, but this is not sure.

Which of ABS-clown-(verbal) you-CTE and ABS-clown-(verbal) you-THM would be correct?

If angry-(verbal) you-AFF this-CTE means ‘this is [an instance of] you being angry’, then
ABS-clown-(verbal) you-CTE should mean "you are an instance of 'being a clown'" / "you are a state of being a clown"

So ABS-clown-(verbal) you-THM is more correct to me

SNT EFF MPL

What are SNT and MPL?

Also why did you not consider the EMBodiment and a new ConTEnt that I've proposed? (just curious of the reason)

TODO: how about implementing inverse Case accessors, too? This would help shave off a significant amount of coantonymic Cases. Observe that V’V forms are not restricted to monophthongal Vs, and instances of )

Implementing case accessors seems worth, but I think it's better to do it by inserting an h for reverse accessors. This is simpler and cleaner (and personally I don't mind [ç~j̊] and [ʍ], which I expect to appear when inserting h in a diphthong)

The lexicon is redesigned so that every root has at most six definitions (2 Designations × 3 Stems); it is not required that every root have all of its Stems defined.

If so, we should enforce Stem 0 as being the generic Stem, not Stem 1.

(TODO(anybody): discuss removing Designation in favour of a large pool (5–8) of Stems).

I would prefer having a semantically-productive Designation rather than an opaque one, but allow to describe "reasonable gloss" in the dictionary.

Having an opaque Designation only makes sense, IMHO, if it is really irregular (not just "domesticated version of" or "in a quasi-permanent context"), and is not just a lexical kitchen sink

For example, the FML of -Ň- (from the v3.1 list) are good to me, but those of ith2k11's -ŇKY- are not

[este-rs]

Okay, one thing at a time:

Technically not exact,

In fact, this is technically exact, as any root can have all of these 24 cells defined separately. We already see examples of this in the lexicon, like -JJ-, which defines all combinations of Specification and Stem (yielding 12 cells; -JJ- exhibits no Formal Stems). ‘Grid’ here is meant more as the principal matrix about which root semantics is vowen. You could argue that I should’ve said 32—after all, -S- defines its ‘Stem 0’ (I still find it hilarious that JQ tends to put Stem 0 in quotes) explicite, so so should be expected of other roots, implicite.

if the definitions of the 4 Specifications count as "rules", then why not use those rules for Stem 1 also (if they are productive enough..)

Yes, the point of the following section is to disprove that assumption.

I would change the wording to "This root exemplifies several observations that can be worded as"

Meh. The word ‘law’ is crucial here, as a TNIL learner—or designer—wishes to regularize the lexicon that he’s to deal with. Of course, in this case, ‘law’-RPV, as the following paragraphs attempt to dispel the notions there introduced.

it can be deduced that:

This is not correct. Function is not a lexical Category, so it can’t influence which lexeme we pick. Function describes our perspective on a given event: Dynamic Function ‘indicates that the verb refers to a tangible or physical act or cause-and-effect event’ (http://www.ithkuil.net/05_verbs_1.html#Sec5o1o2), so it’s more about recasting an event as a change-of-state. I believe the instances of ‘(to be)’ present in definitions where there already is a semicolon and a verbal definition to be in error.

Having "to be a state of" is harmful because it adds a layer between the base meaning of the root, and the argument structure that Cases can access. The core slots/roles of the base root are not usable anymore, only remains the mostly ininteresting "THM is a state of ..."

Indeed. RFC3 attempts to solve this by inserting the ‘the state/action of X’ing’ meaning in the newly-allocated Contentive Case, thus at no times obscuring the underlying case roles.

Sounds a good name to me, but

‘Format’ is appropriate because:

1. it alludes to 2011 Ithkuil Format, which was distinct from Case and characterized the connection between the incorporated root and the main root, just like in our case (between the root and that semantic role within the root which we want);
2. JQ’s 2020 ‘Format’ is nothing but Case of the incorporated root. Not a new concept.

If angry-(verbal) you-AFF this-CTE means ‘this is [an instance of] you being angry’, then
ABS-clown-(verbal) you-CTE should mean "you are an instance of 'being a clown'" / "you are a state of being a clown"

I don’t follow this logic. Contentive is a no-op Case; it refers to that which the root refers to itself. There are two cases (not Cases) under consideration here:

1. Clause Cases are the verbal root’s case roles: SNT-clown-(verbal) you-AFF has you in the Affective Case—you are affected by somebody doing clown stuff—and there is emphasis on the Absolutive party that’s involved in the event of clowning. (Somewhat similar to 2011 Ithkuil, though useless.) CTE would in turn refer to clown.
2. Clause Cases are case roles of the verbal formative at large: SNT-clown-(verbal) you-AFF means that you are affected by the physical expression/manifestation of a clown(ing). CTE would, in turn, refer to SNT-clown.

On second thoughts, (2) is much more sensible; I will push a correction soon.

What are SNT and MPL?

Essential, Implicative. The new Cases are explained in order below the Case table.

Also why did you not consider the EMBodiment and a new ConTEnt that I've proposed? (just curious of the reason)

I believe that’s what Methodic and Essential are.

Implementing case accessors seems worth, but I think it's better to do it by inserting an h for reverse accessors.

Good idea. (-w/y- would work too.) Worth noting is that some case pairs have both of their members equally common, so such a design would need to be irregular from the get-go.

If so, we should enforce Stem 0 as being the generic Stem, not Stem 1.

Stellar idea. I’ll edit that in.

For example, the FML of -Ň- (from the v3.1 list) are good to me, but those of ith2k11's -ŇKY- are not

Agreed. The point is that whenever we want to group related meanings, we should be able to do that without hassle, and without the need to conform to any super-specific, number-precise table scheme. -Ň- could be transported to Freetnil as one root, six roots, or maybe two roots (e.g., you could group Informal 3, Formal 1, and Formal 3 together).

I want there to be ways to use bags-of-words when—and only when—there’s words to ‘bag together’ in the first place. JQ’s ‘eh, whatever, guys, let’s just force every root to contain 6 distinct meanings—every root!’ sort of design, to me, is odious.

[mklcp]

In fact, this is technically exact

"practically" seems more appropriate than "technically" then
Technically, I don't think it's correct to use "technically" there, because, while it could be done, it is never done so.
This makes Specification even more weird: for Stem 1, it is lexical, but for Stems 2 and 3, you're expected to deduce them out of nowhere..

I still find it hilarious that JQ tends to put Stem 0 in quotes

Yeah this is ridiculous.

Yes, the point of the following section is to disprove that assumption.

I understood, this was a rant against Specification

Meh. The word ‘law’ is crucial here, as a TNIL learner—or designer—wishes to regularize the lexicon that he’s to deal with.

Makes sense. But I was describing Specification's flawed point of view, for which those observations do not make real laws in the end. Even if they should.

This is not correct. Function is not a lexical Category, so it can’t influence which lexeme we pick. Function describes our perspective on a given event

I know, and I agree with you. But this hypothesis makes sense, as Ilmen noted.

Indeed. RFC3 attempts to solve this by inserting the ‘the state/action of X’ing’ meaning in the newly-allocated Contentive Case, thus at no times obscuring the underlying case roles.

On that particular point, I don't think having "state/action" as the default semantic is good. I would prefer something vaguer, like an equivalent to lojban's {suhu}

If angry-(verbal) you-AFF this-CTE means ‘this is [an instance of] you being angry’, then
ABS-clown-(verbal) you-CTE should mean "you are an instance of 'being a clown'" / "you are a state of being a clown"

I don’t follow this logic. Contentive is a no-op Case; it refers to that which the root refers to itself.

Seems that we agree that (2) should be chosen. But I don't understand why you don't understand that ABS-clown-(verbal) you-CTE should then mean "you are an instance of 'being a clown'" / "you are a state of being a clown"

here, clause-case CTE is a hook to the role-case of the verb. As a noun,

CTE-clown-(noun)

means

that which is a state of sth being a clown
that sth is a clown

and if you use CTE as a clause-case, like in

ABS-clown-(verbal) you-CTE

then "you" fill the slot of that is referred by CTE-clown-(noun),

i.e.,

ABS-clown-(verbal) you-CTE

means

you are that sth is a clown
you are "an instance of sth being a clown

(NB: I've used `-(noun) to mean any clause-case)

I believe that’s what Methodic and Essential are.

As we discussed, for some concepts like "write", you can distinguish several nuances:

1. the content of the inscription
2. the glyphs written
3. the surface on which the inscription is written
4. the act of writting

 

\4. is done by Methodic
\3. is done by Material

I don't see how Essential fits 1.

It can be argued that ink, for example, is as much a medium as the canvas where a glyph is written. On a second thought, it might not be worth to allocate a new case for 2., but then should Material be expanded to mean both 2. and 3. (for roots like "write"?)

Good idea. (-w/y- would work too.) Worth noting is that some case pairs have both of their members equally common, so such a design would need to be irregular from the get-go.

Yes, we should list up those.

I want there to be ways to use bags-of-words when—and only when—there’s words to ‘bag together’ in the first place. JQ’s ‘eh, whatever, guys, let’s just force every root to contain 6 distinct meanings—every root!’ sort of design, to me, is odious.

Can't agree more, this is damn ugly

[porpoiseless]

Seconding "Make stem 0 default".

These are great ideas. I'm a little worried about case-based derivation, however. How will it be when there is an incorporation, and a word contains:

1. it's own case in the sentence,
2. the case-accessor-stem of its main root,
3. the case-accessor-stem of its incorp. root,
4. the case/format of incorporation

I'll need to play around with some examples... but in the end a scheme like this will make writing the dictionary so much easier!

[este-rs]

@MELOPEE:

Technically, I don't think it's correct to use "technically" there, because, while it could be done, it is never done so.

This makes Specification even more weird: for Stem 1, it is lexical, but for Stems 2 and 3, you're expected to deduce them out of nowhere..

The same follows for 2011, at least partly. What I mean—first and foremost—is two things:

1. 2020’s ‘24-cell grid’ = there are 24 lexemes to choose from.
2. There are cases where these lexemes are systematized—e.g., joint Specification variants, correlated Stems across Designation, etc.

Technically, I should’ve used a word that’s somewhere between ‘practically’ and ‘technically’. Woe is me =)

But this hypothesis makes sense, as Ilmen noted.

Eh, the fact that we have to argue about this shows that this is not how it ought to have been designed.

On that particular point, I don't think having "state/action" as the default semantic is good. I would prefer something vaguer, like an equivalent to lojban's {suhu}

Sorry, didn’t make myself clear:

1. Root stems should principally refer to states/events/propositions (hold your criticism here—I’ll explain myself in a little). I believe that you’ll agree with me when I say that it’s far superior to start with events (spatio-temporal occurences), which are equipped with case roles—which we can extract at any time—than with the objects of those events.
2. The Thematic Case shall always refer to the theme of a base meaning; the Contentive Case shall always refer to the base meaning itself, without any modification. Thus, if the matrix predicate means ‘state of knowing’/‘to know’ (the basic meaning of such a root—2020’s -ŢT- Stem 1), THM nouns are identified as the thing that’s known. and CTE nouns—as the knowing itself. And if the matrix predicate bears a THM instead of a CTE—remember, CTE is a no-op—then a noun declined for CTE shall refer to the knowledge attained, and THM—who knows—possibly the theme of that which is known (i.e., when we take it as an action/state/proposition).
3. In this sense, CTE can be translated as ‘the state/act of X’ing’ iff the predicate means ‘the state/act of X’ing’ (which it should do); however, if the predicate is not a state/act, CTE will point to that derived meaning—not that of the state/act.

And as for states, events, propositions: I agree. These should be separate. Currently, there’s no mechanism for this, so I assume that stems/Stems refer to occurrences (spatio-temporally delineated tangible events or intangible states with some metric by which we may judge where these states start and end*)—with universal propositions such as ‘2 + 2 = 4’ corresponding to universal occurrences, i.e., those which span all of spacetime (and possibly all possible worlds, if/once we include them in our model)—from which ‘raw propositions’ may be derived by substituting the occurrence template (i.e., the extensional definition of all occurrences, from which we typically pick one, which we then signify through employing it in a sentence, just like ‘I love you’ uses the generic ‘love’ (a verb), but still points to a specific occurrence—in particular, one which concerns myself and yourself) for ‘X’ in the expression ‘there exists an occurrence of X’. (There may be other ways to do this; this belongs in a separate thread.)

[clause application discussion]

CTE-clown shall always mean ‘the state/action of being a clown’ (or, alternatively, ‘the proposition: to be a clown’; see above). If we place this in verb position, we get a handy way to attach case roles (which narrow down which occurrence of the abstract sense of the root we’re trying to signify). This isn’t unlike JQ’s design, which allows Cases to attach to already Case-equipped formatives—as if they were the predicates, local to their surroundings. (Framed formatives are just one facet; Case-Scope is what allows even funkier mash-ups. In general, the overall structure of a sentence is a tree, and each node can be nominal or verbal, depending on which way we’re looking at it from: if from the bottom looking up—as subordinate case roles—verbal; if from the top looking down—as the formative which governs a case role—nominal. In this sense, Framed formatives are just a syntactic tool for replacing the default target of Cases. And for relative clauses, but alas, I shan’t speak of them now.) So CTE-clown-(verbal) may be rendered in English as ‘there is clowning going on’, while CTE-clown-(nominal)—‘a state of clowning’. No asymmetry here, unlike JQ’s shenanigans. Same with IND-clown: as a verbal formative, it would mean ‘there is a clown’; as a nominal—‘a clown’. (The only problem here is that it doesn’t make sense to attach case roles onto this extracted IND party, as humans aren’t states, just like you can’t predicatize a constant on a whim if it isn’t predicate-like itself. This amounts to the fact that THM-degree.of.size doesn’t make sense with Cases attached to it, while THM-know does, because facts of knowledge are states rather than objects in the world (objects aren’t knowable, only how they manifest is)).

1. the content of the inscription
2. the glyphs written
3. the surface on which the inscription is written
4. the act of writting

1. Thematic.
2. Resultative, or Methodic then Resultative.
3. Material.
4. That’s the base meaning of the root, so you’d use CTE—the no-op case—if you needed to.

@porpoiseless:

Seconding "Make stem 0 default".

(615bc07).

I'm a little worried about case-based derivation, however […]

Truly, this could get stuffy. But TNIL’s formative structure is potent enough already. With little modification, we can make this work—take a look:

1. Simple formatives: (Format vowel)–consonant root–Stem, Function vowel–Configuration consonant–Case vowel
2. Complex formatives: (incorporated Format vowel)–incorporated Stem consonant–Format vowel–consonant root and etc.

[mklcp]

@porpoiseless

Yeah I thought about that too, but if Specification is better as Case-under-the-hood, let it be?

@uakci

Eh, the fact that we have to argue about this shows that this is not how it ought to have been designed.

Probably the most interesting comment on this peculiar point. It being meta is also not a good sign that the content behind is sound.

Sorry, didn’t make myself clear:

I agree with everything here
Just a quick question
Would CTE be the role-case of "love" in "my love for you"?

just like ‘I love you’ uses the generic ‘love’ (a verb), but still points to a specific occurrence

Yes, assertion is pretty similar to reference, actually.

This isn’t unlike JQ’s design, which allows Cases to attach to already Case-equipped formatives

Ah, good point.
Should the Possessives and Appositives bind to a role/slot of the verb?
Case-scope is still there...

CTE-clown shall always mean ‘the state/action of being a clown’

Yeah so CTE-clown-(verb) you-ABS should mean "you are a state of clowning".

This amounts to the fact that THM-degree.of.size doesn’t make sense with Cases attached to it, while THM-know does, because facts of knowledge are states rather than objects in the world (objects aren’t knowable, only how they manifest is)).

No, I would use degree.of.size the following way:

degree.of.size-(verb) you-ABS intense-THM

to mean

< you are big

thus

THM-degree.of.size

would mean

< a certain degree of size / a certain size

Resultative, or Methodic then Resultative.

I don't agree. I would use a reverse PDC

Truly, this could get stuffy. But TNIL’s formative structure is potent enough already. With little modification, we can make this work—take a look:

Let me remind you of the following idea: by default, if clause-case and case-accessor (new format) of the main root are the same, one of them could be elided.
Same for case of incorporation and case-accessor of incorporated root.

[este-rs]

Would CTE be the role-case of "love" in "my love for you"?

No: the CTE for ‘my love for you’ would be ‘my love for you’.

Should the Possessives and Appositives bind to a role/slot of the verb?

Case-scope is still there...

Yes—I believe they’re a sort of case role.

No, I would use degree.of.size the following way […]

That’s exactly how I’d use it, too. (???) Perhaps you misunderstood what I’d said here:

This amounts to the fact that THM-degree.of.size doesn’t make sense with Cases attached to it, while THM-know does,

through which I was trying to say that THM-degree.of.size isn’t a proposition/event/state in and of itself, and hence can’t govern any cases without spawning nonsense (what does it mean to be the IND of a degree of size?)—maybe except for the Possessive Cases series. But CTE-degree.of.size—sure.

I don't agree. I would use a reverse PDC

You’re free to elaborate on the case design. I’m not dissatisfied with my design, but if you are, you’re welcome to make changes to the table and stuff.

Let me remind you of the following idea: by default, if clause-case and case-accessor (new format) of the main root are the same, one of them could be elided.

I disagree—this wouldn’t come into play too often: CTE Format + FUN Case is much more sensible than FUN Format + FUN Case (unless you’re trying to say something like ‘I’m teasing you in the way that you ridicule me’, for similes—but that’s not too common either). And CTE would be implied in many places already (the pre-root vowel having it as the omissible default, for example).

Same for case of incorporation and case-accessor of incorporated root.

How to implement root incorporation is a whole another discussion… :)

[Ntsekees]

As I understand it, the CTE case suggested by Uakci is the same thing as my original vision for the REL (Relative) case, before JQ decided to make it having adnominal scope by default and always open a subordinate clause by default. I've also come to wonder whether it's not redundant to the Functive case, concerned with so-called “manners” of events. In English, “she danced in a beautiful manner” is synonymous with “she danced beautifully”, which can also be reworded as “here dancing was beautiful”. Here, the concept of “manner” vanish, and we merely qualify the event of dancing with the property of being beautiful. In which case, “dance-V she-IND beautiful-CTE/FUN” may be the same thing.