Zompist Bboard Again

Would it be reasonable to have unpredictable suffixes in a proto-language for simple/basic verbs? So, in my P conlang it is going to have three possible suffixes for verbnouns (like Celtic): these are -ed, -eb, -ia (/ja/). These perform the functions of infinitive, present/active participle, and gerund. The idea being that these originally stem from the proto-PQL gerunds and participles; due to their similarities they merged during the Proto-PQ stage (a stage I view as being similar to Proto-Celtic).

What I define a simple or basic verb is one which comes from a verbal root in Proto-PQL - its verbal stem is the root: so the root √TAL gives the verbal stem tal- to which suffixes for tense, aspect, mood are attached. The suffixes for gerunds need to be -ikʷā, -itā, and -jā in order to give -eb, -ed, and -ia later in P, but would it be reasonable for there to be three associated with basic verbs so that there is no discernible pattern for which form a basic verb would have had?

Short answer: yes. Unpredictable verbal nouns are quite typical for the older stages of IE languages; you mention Celtic. Only in IE the pathway was the other way round - verbal nouns are the older formation. Gerunds, infinitives, etc., have been formed later in the history of the IE laguages based on case forms of verbal nouns.

hwhatting wrote: ↑Mon Jun 19, 2023 7:10 am Short answer: yes. Unpredictable verbal nouns are quite typical for the older stages of IE languages; you mention Celtic. Only in IE the pathway was the other way round - verbal nouns are the older formation. Gerunds, infinitives, etc., have been formed later in the history of the IE laguages based on case forms of verbal nouns.

SO there's no reason then why a similar verb of the same type (i.e. "basic") couldn't have a different form, e.g. tal-itā next to cal-ikʷā? I'm glad. I'm not trying to do the IE thing too closely; the proto-lang has infinitives and gerunds, etc so one could speculate that it could go further back but I'm not concerned with it - it has to start somewhere. Thanks.

I've been working or reworking a phonology with apocope and syncope, and I was considering how it might be interesting to add ditropic case clitics into it. A ditropic clitic is a clitic that attach one way phonologically but another way syntactically. For example, Belep has case clitics which attach to the word before the noun phrase whose case is being marked:

ja=me pan=e teâmaa
1PL.INCL.SUBJ=IRR go.TV=DAT high.chief
"We will go to the chieftain"

Note that dative =e is attached to the preceding verb despite the fact it governs the noun teâmaa 'high chief' that follows it. These are not applicatives because the enclitic can be hosted by almost any preceding word, including other arguments (NPs).

Anyway, some languages with apocope and syncope seem to have enclitics which shield or block apocope and otherwise behave as part of the word for vowel deletion processes. Aguaruna is such a language, at least according to the grammar of Simon Overall. It has case enclitics which occur at the end of NPs, and which block apocope of the host's final vowel (since the final vowel isn't final anymore after attachment of the enclitic). For example, namakan = namaka-na fish-ACC, with the final vowel of namaka preserved in the accusative.

So if you combine regular vowel deletion processes with ditropic clitics, which are both independently attested in natlangs, you could get some interesting alternations in the preceding word depending on the case marking of the following NP. For example, if we assume apocope followed by a productive syncope rule that deletes the vowel from a second light syllable (CVCV), then you can get alternations like:

kisin <-- form after apocope
kisina
see
"he/she sees"

kisnal tasaq <-- form after apocope and syncope
kisina=li tasaq
see-ERG man
"the man sees him/her/it"

If the mutating word is the verb and you're just comparing VO and VS, it's almost like an inverse or Phillipine voice system, but of course it stops looking like that if you start inserting other elements. E.g. in a VOS clause, the ergative case marker would attach to the O, not to the initial verb.

I don't know how feasible this is. I don't know of a language where it happens, but ditropic clitics are not that common anyway so it could just be a small sample issue.

The other interesting but maybe unattested thing you could do is something like the contiguity marking of Macro-Je and other Amazonian languages via a phrase-final apocope rule. Macro-Je languages often show alternations in prefixes depending on whether a verb, preposition etc. is directly preceded by its argument or not (thus "contiguity marking"), but in the Macro-Je case the alternation is morphological. If you had a rule of apocope that only applied to the phonological phrase final word, for some definition of the phonological phrase, then you could achieve something like case marking by presence or absence of final vowels. Compare:

kisin <-- form after apocope
kisina
see

kisina tasaq <-- no apocope because VO forms a phrase and the V is not final
kisina tasaq
see man.OBJECT
(he/she/it) sees the man

kisin tasaq <-- apocope because VS does not form a phrase for the purpose of the rule
kisina tasaq
see man.SUBJECT
the man sees (him/her/it)

This is basically a linker like the ubiquitous Tagalog ng, only it's just an underlying lexical specified vowel that gets deleted otherwise. I'm less convinced by the second idea than the first, which I am seriously considering for a conlang.

chris_notts wrote: ↑Tue Jul 18, 2023 4:54 pm I don't know how feasible this is. I don't know of a language where it happens, but ditropic clitics are not that common anyway so it could just be a small sample issue.

As long as the syncope stays transparent, it would probably be somewhat stable. After all, Nishnaabemwin exists. But I’d expect it would rapidly disintegrate if the syncope is obscured, as happened in e.g. Old Irish.

This is basically a linker like the ubiquitous Tagalog ng, only it's just an underlying lexical specified vowel that gets deleted otherwise. I'm less convinced by the second idea than the first, which I am seriously considering for a conlang.

On the contrary, I find the second idea a lot more convincing.

I think I understood the part I've emphasized and underlined, but I want to ask and be certain...

chris_notts wrote: ↑Tue Jul 18, 2023 4:54 pm I've been working or reworking a phonology with apocope and syncope, and I was considering how it might be interesting to add ditropic case clitics into it. A ditropic clitic is a clitic that attach one way phonologically but another way syntactically. For example, Belep has case clitics which attach to the word before the noun phrase whose case is being marked:

ja=me pan=e teâmaa
1PL.INCL.SUBJ=IRR go.TV=DAT high.chief
"We will go to the chieftain"

Note that dative =e is attached to the preceding verb despite the fact it governs the noun teâmaa 'high chief' that follows it. These are not applicatives because the enclitic can be hosted by almost any preceding word, including other arguments (NPs).

so...if I remember the Syntax Construction Kit and various glossaries, that means that the statement can be parsed, for example, as
(we) ((go) to) (high chief)

...so the final <e> is
(we) ((go) to) (<e> high chief) ....that is, it is a part of the chieftan's portion of the statement.

...even though the final <e> is pronounced as part of
(pan=e = (go) to)

Yes?

keenir wrote: ↑Tue Jul 18, 2023 11:41 pm I think I understood the part I've emphasized and underlined, but I want to ask and be certain...

chris_notts wrote: ↑Tue Jul 18, 2023 4:54 pm I've been working or reworking a phonology with apocope and syncope, and I was considering how it might be interesting to add ditropic case clitics into it. A ditropic clitic is a clitic that attach one way phonologically but another way syntactically. For example, Belep has case clitics which attach to the word before the noun phrase whose case is being marked:

ja=me pan=e teâmaa
1PL.INCL.SUBJ=IRR go.TV=DAT high.chief
"We will go to the chieftain"

Note that dative =e is attached to the preceding verb despite the fact it governs the noun teâmaa 'high chief' that follows it. These are not applicatives because the enclitic can be hosted by almost any preceding word, including other arguments (NPs).

so...if I remember the Syntax Construction Kit and various glossaries, that means that the statement can be parsed, for example, as
(we) ((go) to) (high chief)

...so the final <e> is
(we) ((go) to) (<e> high chief) ....that is, it is a part of the chieftan's portion of the statement.

...even though the final <e> is pronounced as part of
(pan=e = (go) to)

Yes?

Almost, but not quite. Syntactically, this statement is:

[we] [IRR go] [to high.chief]

But phonologically, it’s:

we=IRR go=to high.chief

So the clitic associates ‘the wrong way round’.

(Some English clitics behave similarly: in a sentence like ‘I’ll do it’, there is no syntactic constituent corresponding to the two words ‘I will’. But I’m not sure if this counts as being properly ditropic or not.)

bradrn wrote: ↑Tue Jul 18, 2023 11:44 pm

keenir wrote: ↑Tue Jul 18, 2023 11:41 pm I think I understood the part I've emphasized and underlined, but I want to ask and be certain...

chris_notts wrote: ↑Tue Jul 18, 2023 4:54 pm I've been working or reworking a phonology with apocope and syncope, and I was considering how it might be interesting to add ditropic case clitics into it. A ditropic clitic is a clitic that attach one way phonologically but another way syntactically. For example, Belep has case clitics which attach to the word before the noun phrase whose case is being marked:

ja=me pan=e teâmaa
1PL.INCL.SUBJ=IRR go.TV=DAT high.chief
"We will go to the chieftain"

Note that dative =e is attached to the preceding verb despite the fact it governs the noun teâmaa 'high chief' that follows it. These are not applicatives because the enclitic can be hosted by almost any preceding word, including other arguments (NPs).

so...if I remember the Syntax Construction Kit and various glossaries, that means that the statement can be parsed, for example, as
(we) ((go) to) (high chief)

...so the final <e> is
(we) ((go) to) (<e> high chief) ....that is, it is a part of the chieftan's portion of the statement.

...even though the final <e> is pronounced as part of
(pan=e = (go) to)

Yes?

Almost, but not quite. Syntactically, this statement is:
[we] [IRR go] [to high.chief]

But phonologically, it’s:
we=IRR go=to high.chief

(we) (IRR go) (to high chief) vs (we IRR) (go=to) (high chief)

I confess I was staring at this and what I'd guessed above for several minutes, trying to see what the difference was, before I realized that, while I'd put the <e> in with <high chief> in the syntactic line...I'd also put <to> in with <go>, not realizing at the time that, in this instance at least, the <e> = <to>

Was that the almost?

So the clitic associates ‘the wrong way round’.

?Because if it was based on its syntactic role (and nothing more than that), it would have the final <e> as part of < teâmaa> because that means <high.chief>, which is what the <e>'s meaning related to...whereas it if it was adhering strictly to the phonological placement, then the final <e> would have a meaning that had to do with <go to> because the <e> was pronounced as part of the word <pan=e> ?

keenir wrote: ↑Wed Jul 19, 2023 12:00 am I confess I was staring at this and what I'd guessed above for several minutes, trying to see what the difference was, before I realized that, while I'd put the <e> in with <high chief> in the syntactic line...I'd also put <to> in with <go>, not realizing at the time that, in this instance at least, the <e> = <to>

Was that the almost?

So the clitic associates ‘the wrong way round’.

?Because if it was based on its syntactic role (and nothing more than that), it would have the final <e> as part of < teâmaa> because that means <high.chief>, which is what the <e>'s meaning related to...whereas it if it was adhering strictly to the phonological placement, then the final <e> would have a meaning that had to do with <go to> because the <e> was pronounced as part of the word <pan=e> ?

Yes to all of this!

bradrn wrote: ↑Wed Jul 19, 2023 12:27 am

keenir wrote: ↑Wed Jul 19, 2023 12:00 am I confess I was staring at this and what I'd guessed above for several minutes, trying to see what the difference was, before I realized that, while I'd put the <e> in with <high chief> in the syntactic line...I'd also put <to> in with <go>, not realizing at the time that, in this instance at least, the <e> = <to>

Was that the almost?

So the clitic associates ‘the wrong way round’.

?Because if it was based on its syntactic role (and nothing more than that), it would have the final <e> as part of < teâmaa> because that means <high.chief>, which is what the <e>'s meaning related to...whereas it if it was adhering strictly to the phonological placement, then the final <e> would have a meaning that had to do with <go to> because the <e> was pronounced as part of the word <pan=e> ?

Yes to all of this!

Thank you for the help. Very much appreciated.

(I wasn't sure if that was the almost, or if there was something else I missed - knowing me, I didn't discount the possibility)

keenir wrote: ↑Wed Jul 19, 2023 12:31 am (I wasn't sure if that was the almost, or if there was something else I missed - knowing me, I didn't discount the possibility)[/size]

Yes, that was the almost.

bradrn wrote: ↑Wed Jul 19, 2023 12:33 am

keenir wrote: ↑Wed Jul 19, 2023 12:31 am (I wasn't sure if that was the almost, or if there was something else I missed - knowing me, I didn't discount the possibility)[/size]

Yes, that was the almost.

There's a good paper here which discussed some examples and what counts, for the enclitic case anyway:

http://cysouw.de/home/articles_files/cy ... LITICS.pdf

chris_notts wrote: ↑Wed Jul 19, 2023 3:27 am

bradrn wrote: ↑Wed Jul 19, 2023 12:33 am

keenir wrote: ↑Wed Jul 19, 2023 12:31 am (I wasn't sure if that was the almost, or if there was something else I missed - knowing me, I didn't discount the possibility)[/size]

Yes, that was the almost.

There's a good paper here which discussed some examples and what counts, for the enclitic case anyway:

http://cysouw.de/home/articles_files/cy ... LITICS.pdf

I feel sure I’ve seen Cysouw’s name before in connection to clitics, but I don’t think I’ve seen this particular paper before.

(keenir’s issue was something else, though.)

This discussion reminded me of Kusaal , an Oti-Volta language spoken in Ghana and Burkina Faso. It has "invisible clitics" or, as they're called in the grammar I link, "prosodic enclitics", which have no surface realisation of their own, but undo apocope and effects of tone sandhi, e.g. the negation:

Lɩ̀ ka' dʋkɔ́·. "It's not a pot (dʋk_ɔˊ)."
3i NG.be pot·NG
Lɩ̀ ka' nɔ́bɩ̀rɛ·. "It's not a leg (nɔ́bɩ̀r)."
Bà ka' mɔ́lii·. "They are not kobs (mɔ̀lì)."
Lɩ̀ ka' yàarɩmm·. "It's not salt ( yàarɩ̀m)."
Lɩ̀ ká' ò tɩɩmm·. "It's not her medicine (tɩ̀ɩm)."
Lɩ̀ ká' bà da'a·. "It's not their market (dà'a)."

(Examples from p. 26 of the linked grammar)

I don't remember if it has any such clitics that could be described as ditropic, but I never read the entire grammar - I know features of Kusaal mostly because its author, David Eddyshaw, is a regular on Languagehat and often quotes examples and features from Kusaal in linguistic discussions.

hwhatting wrote: ↑Wed Jul 19, 2023 4:47 am This discussion reminded me of Kusaal , an Oti-Volta language spoken in Ghana and Burkina Faso. It has "invisible clitics" or, as they're called in the grammar I link, "prosodic enclitics", which have no surface realisation of their own, but undo apocope and effects of tone sandhi, e.g. the negation:

Lɩ̀ ka' dʋkɔ́·. "It's not a pot (dʋk_ɔˊ)."
3i NG.be pot·NG
Lɩ̀ ka' nɔ́bɩ̀rɛ·. "It's not a leg (nɔ́bɩ̀r)."
Bà ka' mɔ́lii·. "They are not kobs (mɔ̀lì)."
Lɩ̀ ka' yàarɩmm·. "It's not salt ( yàarɩ̀m)."
Lɩ̀ ká' ò tɩɩmm·. "It's not her medicine (tɩ̀ɩm)."
Lɩ̀ ká' bà da'a·. "It's not their market (dà'a)."

(Examples from p. 26 of the linked grammar)

I don't remember if it has any such clitics that could be described as ditropic, but I never read the entire grammar - I know features of Kusaal mostly because its author, David Eddyshaw, is a regular on Languagehat and often quotes examples and features from Kusaal in linguistic discussions.

Very cool! Yes, this is effectively my second option, a linking clitic whose only realisation is undoing apocope and maybe as a result triggering syncope in the penultimate syllable.

keenir wrote: ↑Tue Jul 18, 2023 11:41 pm

chris_notts wrote: ↑Tue Jul 18, 2023 4:54 pm I've been working or reworking a phonology with apocope and syncope, and I was considering how it might be interesting to add ditropic case clitics into it. A ditropic clitic is a clitic that attach one way phonologically but another way syntactically. For example, Belep has case clitics which attach to the word before the noun phrase whose case is being marked:

ja=me pan=e teâmaa
1PL.INCL.SUBJ=IRR go.TV=DAT high.chief
"We will go to the chieftain"

Note that dative =e is attached to the preceding verb despite the fact it governs the noun teâmaa 'high chief' that follows it. These are not applicatives because the enclitic can be hosted by almost any preceding word, including other arguments (NPs).

so...if I remember the Syntax Construction Kit and various glossaries, that means that the statement can be parsed, for example, as
(we) ((go) to) (high chief)

...so the final <e> is
(we) ((go) to) (<e> high chief) ....that is, it is a part of the chieftan's portion of the statement.

...even though the final <e> is pronounced as part of
(pan=e = (go) to)

Since you invoke the SCK I thought I'd comment. First: nice feature! I love when languages get weird.

And yes, the surface structure would be ((go) to). A GG approach would probably derive that from (go (to (high chief))), though God knows what the Chomskyan position would be.

Brad's "I'll" is arguably an example in English; I'd add "wanna" as in "I wanna read that book." (That is, I assume infinitive "to" goes with the following verb. But I'm not 100% sure, I'd have to look at a bunch of variations to check.)

zompist wrote: ↑Wed Jul 19, 2023 6:42 pmSince you invoke the SCK I thought I'd comment.

Please and thank you!

And yes, the surface structure would be ((go) to). A GG approach would probably derive that from (go (to (high chief))), though God knows what the Chomskyan position would be.

I would've thought that it would be ((we) (go)) separate from (go (to (high.chief))) except at the very top of the bifurcations. Though I may be misunderstanding/missing what you meant by "surface structure" (if its different from what was just explained to me, then i definately missed)

Brad's "I'll" is arguably an example in English; I'd add "wanna" as in "I wanna read that book." (That is, I assume infinitive "to" goes with the following verb. But I'm not 100% sure, I'd have to look at a bunch of variations to check.)

Well, its the DAT (dative), so it goes with "high.chief"...but yeah, it makes sense even without the "to/that book" -- "I wanna read" is perfectly acceptible English, at least in the parts of Maryland and North Carolina I've heard people speak in.

keenir wrote: ↑Wed Jul 19, 2023 8:14 pm

And yes, the surface structure would be ((go) to). A GG approach would probably derive that from (go (to (high chief))), though God knows what the Chomskyan position would be.

I would've thought that it would be ((we) (go)) separate from (go (to (high.chief))) except at the very top of the bifurcations. Though I may be misunderstanding/missing what you meant by "surface structure" (if its different from what was just explained to me, then i definately missed)

Surface structure is the final tree for the sentence, after any transformations.

As you say, phonologically the dative is attaching to the previous word.

What would be syntactic structure be? One way is to match the phonology: e.g. pan-e is a constituent ((go) to). But from that description, it sounds like -e attaches to any previous material-- so pan-e is probably not a constituent. (We could test this by using a transformation to move pan-, and see if -e moves with it.)

So I'd expect the syntactic structure is

(we) (go) (to (chief))

I don't think this would be a problem... the phonological component can handle much harder stuff than that.

One point which maybe hasn’t been made sufficiently clear yet is that phonological and grammatical constituents do not in general coincide. There’s no reason why you can’t have one tree structure for syntax, and a different one for phonology. Ditropic clitics are particularly good examples of this, but it’s not hard to invent examples in other languages. For instance:

The man who I saw yesterday had a car with a flat tyre.

Phonologically, the intonational structure is something like this:

[The man] [who I saw yesterday] [had a car] [with a flat tyre]

Whereas the syntactic structure is quite different:

[The man who [I saw yesterday]] [had [a car [with a flat tyre]]]

bradrn wrote: ↑Tue Jul 18, 2023 11:44 pm (Some English clitics behave similarly: in a sentence like ‘I’ll do it’, there is no syntactic constituent corresponding to the two words ‘I will’. But I’m not sure if this counts as being properly ditropic or not.)

At least in English there are many cliticized words which do not join phonologically with the word that they form a syntactic phrase with, but instead with the preceding word. Not just will, but have, would, had, did, to, of, was, were, and so on all do this.