Conlang Random Thread

[Travis B.]

Today I've been thinking about the "areal linguistics" of the region (unnamed) my primary conlang Vrkhazhian is found in. For instance, I would say that the languages of the area have this sort of phonological inventory:

/m n ŋ/
/p b t d k g/ or /p pʰ t tʰ k kʰ/ or /p pʰ t tʰ k kʰ q qʰ/
/f s x/ or /f s ʃ x/
/l r/

Essentially, the languages all have a velar nasal, at minimum a binary distinction in plosives (voicing or aspiration) and the presence of at least /f s x/. Beyond that, the languages often have more than just these consonants and even voiced fricatives.

Morphologically, most are agglutinating and usually have either three or more genders or no genders at all. Regarding case, they either have medium to large case systems or no case at all (i.e. no 2-case or 3-case systems). Of the languages with case, the vocative and equative (to express likeness or equivalence) are common.


Mostly I was thinking about the syllable structure of Vrkhazhian and how the language handles certain homo-organic clusters. I have many words with nasal+stop clusters in the root, such as lumb-, zand- and man̮k- and this poses a problem: the constructed state is indicated by the removal of the gender and case endings, resulting in word-final clusters, which are illegal. This is typically solved by inserting echo vowels between the members of the offending cluster. But the result feels wrong with nasal+stop clusters: lumub, zanad, man̮ak in much the same way it feels to break up a geminate cluster this way.

Instead, I decided they will just keep their gender vowel (i.e. lumb-u, zand-i, man̮k-a) so as to not break the cluster and I needed a justification for this so I decided that these words with nasal+stop clusters come from a substrate language that allowed these clusters word-finally. Thus the behavior is an indication of the words' foreign origins.

Why not just permit homorganic nasal+stop clusters in final position?

[Ahzoh]

Today I've been thinking about the "areal linguistics" of the region (unnamed) my primary conlang Vrkhazhian is found in. For instance, I would say that the languages of the area have this sort of phonological inventory:

/m n ŋ/
/p b t d k g/ or /p pʰ t tʰ k kʰ/ or /p pʰ t tʰ k kʰ q qʰ/
/f s x/ or /f s ʃ x/
/l r/

Essentially, the languages all have a velar nasal, at minimum a binary distinction in plosives (voicing or aspiration) and the presence of at least /f s x/. Beyond that, the languages often have more than just these consonants and even voiced fricatives.

Morphologically, most are agglutinating and usually have either three or more genders or no genders at all. Regarding case, they either have medium to large case systems or no case at all (i.e. no 2-case or 3-case systems). Of the languages with case, the vocative and equative (to express likeness or equivalence) are common.


Mostly I was thinking about the syllable structure of Vrkhazhian and how the language handles certain homo-organic clusters. I have many words with nasal+stop clusters in the root, such as lumb-, zand- and man̮k- and this poses a problem: the constructed state is indicated by the removal of the gender and case endings, resulting in word-final clusters, which are illegal. This is typically solved by inserting echo vowels between the members of the offending cluster. But the result feels wrong with nasal+stop clusters: lumub, zanad, man̮ak in much the same way it feels to break up a geminate cluster this way.

Instead, I decided they will just keep their gender vowel (i.e. lumb-u, zand-i, man̮k-a) so as to not break the cluster and I needed a justification for this so I decided that these words with nasal+stop clusters come from a substrate language that allowed these clusters word-finally. Thus the behavior is an indication of the words' foreign origins.

Why not just permit homorganic nasal+stop clusters in final position?

I just said, final clusters are illegal. No exceptions.

[bradrn]

Today I've been thinking about the "areal linguistics" of the region (unnamed) my primary conlang Vrkhazhian is found in. For instance, I would say that the languages of the area have this sort of phonological inventory:

/m n ŋ/
/p b t d k g/ or /p pʰ t tʰ k kʰ/ or /p pʰ t tʰ k kʰ q qʰ/
/f s x/ or /f s ʃ x/
/l r/

Essentially, the languages all have a velar nasal, at minimum a binary distinction in plosives (voicing or aspiration) and the presence of at least /f s x/. Beyond that, the languages often have more than just these consonants and even voiced fricatives.

Morphologically, most are agglutinating and usually have either three or more genders or no genders at all. Regarding case, they either have medium to large case systems or no case at all (i.e. no 2-case or 3-case systems). Of the languages with case, the vocative and equative (to express likeness or equivalence) are common.


Mostly I was thinking about the syllable structure of Vrkhazhian and how the language handles certain homo-organic clusters. I have many words with nasal+stop clusters in the root, such as lumb-, zand- and man̮k- and this poses a problem: the constructed state is indicated by the removal of the gender and case endings, resulting in word-final clusters, which are illegal. This is typically solved by inserting echo vowels between the members of the offending cluster. But the result feels wrong with nasal+stop clusters: lumub, zanad, man̮ak in much the same way it feels to break up a geminate cluster this way.

Instead, I decided they will just keep their gender vowel (i.e. lumb-u, zand-i, man̮k-a) so as to not break the cluster and I needed a justification for this so I decided that these words with nasal+stop clusters come from a substrate language that allowed these clusters word-finally. Thus the behavior is an indication of the words' foreign origins.

Why not just permit homorganic nasal+stop clusters in final position?

I just said, final clusters are illegal. No exceptions.

I think Travis was saying that you could make an exception for nasal+stop clusters. This is very common in languages which otherwise lack clusters — which is why nasal+stop sequences are so often analysed as single phonemes (‘prenasalised stops’).

[ophois]

Two trains of thought I had during a run, each resulting in some weird but hopefully not unnaturalistic morphosyntax.

The first started from the verb 'to know', which I thought it might be fun to reinterpret as ditransitive in a conlang. So syntactic (I) know (he likes ducks) would be reinterpreted as (I) know (he) (likes ducks) where (he) is an indirect object and (likes ducks) in a nominalised form is the direct object. (c.f. (I) gave (him) (the book) in English). However, (I) know (he likes ducks) is a cardinal example of a sentence that uses a clausal object. Perhaps all such sentences are treated in this way, such that I think we bought icecream is syntactically (I) think (we) (bought icecream).

The second, and more interesting IMO, concerns existentials. In English, a 'gnomic' existential is often used to express the fact that something exists with present relevance (There's a duck, There's someone watching us). If I had to guess, the semantic shift went direct locative (as in 'right there') > relevant existential > gnomic but it feels naturalistic enough to reverse it based on Grice's maxim of relation (so A duck exists (in general) > A duck exists (with present relevance)).

In English (again), at least IMD, we can also use the copula to form such a gnomic (I am, Ducks are) but this use is atypical. The copula, almost by definition, requires two arguments, but there are various ways to knock out an argument. One of these is the middle voice which can be analysed as demoting O and promoting A (even though A is already the primary argument), at least in its prototypical reflexive meaning (This, incidentally, makes it the best equivalent to a nominative-absolutive version of the antipassive). Anyway, we can apply the middle voice to the copula and get a gnomic existential (per Wikipedia, some have analysed the intransitive uses of English labile verbs as actually middle voice anyway, making 'I am' already an example of this strategy). All this to set up the fact instead of 'it rains' with a dummy pronoun, in this hypothetical lang you can say 'rain COP-MID' (i.e. there is rain, rain is).

This might just be incoherent rambling, but it makes sense to me.

[bradrn]

The first started from the verb 'to know', which I thought it might be fun to reinterpret as ditransitive in a conlang. So syntactic (I) know (he likes ducks) would be reinterpreted as (I) know (he) (likes ducks) where (he) is an indirect object and (likes ducks) in a nominalised form is the direct object. (c.f. (I) gave (him) (the book) in English). However, (I) know (he likes ducks) is a cardinal example of a sentence that uses a clausal object. Perhaps all such sentences are treated in this way, such that I think we bought icecream is syntactically (I) think (we) (bought icecream).

Reminds me of raising-to-object.

The second, and more interesting IMO, concerns existentials. In English, a 'gnomic' existential is often used to express the fact that something exists with present relevance (There's a duck, There's someone watching us). If I had to guess, the semantic shift went direct locative (as in 'right there') > relevant existential > gnomic but it feels naturalistic enough to reverse it based on Grice's maxim of relation (so A duck exists (in general) > A duck exists (with present relevance)).

Several comments:

• Grammaticalisation shifts are generally not reversible. They almost invariably go from the more concrete to the more abstract, not the other way around.
• Beware that Grice’s maxims are not at all universal: they apply well to Western culture, but not so well elsewhere. (Wierzbicka’s Cross-Cultural Pragmatics is a good read.)
• For the same reason I doubt that Grice’s maxims are required to justify grammaticalisation processes. The World Lexicon of Grammaticalisation lists examples of ‘locative copula > exist’ from English, ǁAni, Swahili and Nubi, so it’s not restricted to any one culture.

The copula, almost by definition, requires two arguments, but there are various ways to knock out an argument.

Note that many languages forbid valency-changing operations on the copula — e.g. in English you can’t passivise it.

One of these is the middle voice which can be analysed as demoting O and promoting A (even though A is already the primary argument), at least in its prototypical reflexive meaning

This describes an antipassive, not a middle voice. (And note that there’s plenty of accusative languages with an antipassive). To me, a middle voice neither demotes O nor promotes A: it deletes one argument completely, while the other argument remains with its usual marking. Its other key feature is that some monovalent verbs obligatorily take the middle voice (Inglese 2021).

Also, it’s difficult to say that the ‘prototypical meaning’ of a middle voice is reflexive. On the contrary, Inglese finds that an anticausative meaning is most central to the construction.

(per Wikipedia, some have analysed the intransitive uses of English labile verbs as actually middle voice anyway, making 'I am' already an example of this strategy)

Some may have called it that, yes, but it’s completely different in very many ways to the kind of constructions which are more usually called ‘middle’. (Again refer to Inglese… it’s really an excellent and fascinating paper.)

[Travis B.]

Why not just permit homorganic nasal+stop clusters in final position?

I just said, final clusters are illegal. No exceptions.

I think Travis was saying that you could make an exception for nasal+stop clusters. This is very common in languages which otherwise lack clusters — which is why nasal+stop sequences are so often analysed as single phonemes (‘prenasalised stops’).

Exactly. If you feel bad about making an exception for nasal-stop clusters, analyze them as prenasalized stops ─ and there is no problem with also forbidding prenasalized stops initially (e.g. Old and Middle Japanese had prenasalized stop phonemes, which were forbidden in initial position for native lexemes).

[ophois]

The second, and more interesting IMO, concerns existentials. In English, a 'gnomic' existential is often used to express the fact that something exists with present relevance (There's a duck, There's someone watching us). If I had to guess, the semantic shift went direct locative (as in 'right there') > relevant existential > gnomic but it feels naturalistic enough to reverse it based on Grice's maxim of relation (so A duck exists (in general) > A duck exists (with present relevance)).

Several comments:

• Grammaticalisation shifts are generally not reversible. They almost invariably go from the more concrete to the more abstract, not the other way around.
• Beware that Grice’s maxims are not at all universal: they apply well to Western culture, but not so well elsewhere. (Wierzbicka’s Cross-Cultural Pragmatics is a good read.)
• For the same reason I doubt that Grice’s maxims are required to justify grammaticalisation processes. The World Lexicon of Grammaticalisation lists examples of ‘locative copula > exist’ from English, ǁAni, Swahili and Nubi, so it’s not restricted to any one culture.

Thanks for the corrections, I'm a bit of noob, but it's good to know I can simply steal this from English(, Swahili, etc.).

The copula, almost by definition, requires two arguments, but there are various ways to knock out an argument.

Note that many languages forbid valency-changing operations on the copula — e.g. in English you can’t passivise it.

Many, but not all.

One of these is the middle voice which can be analysed as demoting O and promoting A (even though A is already the primary argument), at least in its prototypical reflexive meaning

This describes an antipassive, not a middle voice. (And note that there’s plenty of accusative languages with an antipassive). To me, a middle voice neither demotes O nor promotes A: it deletes one argument completely, while the other argument remains with its usual marking. Its other key feature is that some monovalent verbs obligatorily take the middle voice (Inglese 2021).

Also, it’s difficult to say that the ‘prototypical meaning’ of a middle voice is reflexive. On the contrary, Inglese finds that an anticausative meaning is most central to the construction.

(per Wikipedia, some have analysed the intransitive uses of English labile verbs as actually middle voice anyway, making 'I am' already an example of this strategy)

Some may have called it that, yes, but it’s completely different in very many ways to the kind of constructions which are more usually called ‘middle’. (Again refer to Inglese… it’s really an excellent and fascinating paper.)

It seems this mistake of mine comes from a misintepretation of middle voice. I'll make sure to read Inglese to fully grok what it is. Anyway, the anticausative meaning more fully aligns with my original thoughts, which had e.g. 'rain' acting more patientive than agentive syntacto-semantically (or however you term it).

[bradrn]

The second, and more interesting IMO, concerns existentials. In English, a 'gnomic' existential is often used to express the fact that something exists with present relevance (There's a duck, There's someone watching us). If I had to guess, the semantic shift went direct locative (as in 'right there') > relevant existential > gnomic but it feels naturalistic enough to reverse it based on Grice's maxim of relation (so A duck exists (in general) > A duck exists (with present relevance)).

Several comments:

• Grammaticalisation shifts are generally not reversible. They almost invariably go from the more concrete to the more abstract, not the other way around.
• Beware that Grice’s maxims are not at all universal: they apply well to Western culture, but not so well elsewhere. (Wierzbicka’s Cross-Cultural Pragmatics is a good read.)
• For the same reason I doubt that Grice’s maxims are required to justify grammaticalisation processes. The World Lexicon of Grammaticalisation lists examples of ‘locative copula > exist’ from English, ǁAni, Swahili and Nubi, so it’s not restricted to any one culture.

Thanks for the corrections, I'm a bit of noob, but it's good to know I can simply steal this from English(, Swahili, etc.).

One other thing I should mention: not all of these constructions are precisely the same as the English one, because the other languages here have a dedicated locative verb meaning ‘to be at’. Thus, the Swahili verb -ko means ‘to be at’ with a locative complement, or ‘to exist’ when no complement is specified. In Nubi, the verb fí is transitive meaning ‘to be at’, but in a topic construction it’s existential. And so on.

(It doesn’t even have to be a copular verb: in French the equivalent construction is il y a X, literally meaning ‘it has X there’.)

Anyway, the anticausative meaning more fully aligns with my original thoughts, which had e.g. 'rain' acting more patientive than agentive syntacto-semantically (or however you term it).

That makes no sense to me. The arguments of a copula aren’t agents or patients in the first place — those descriptions don’t fit with their actual semantic roles in a copulative clause. (Which is one reason why, cross-linguistically, copulae often differ from normal verbs in their behaviour, especially when it comes to transitivity.) You can’t apply an anticausative transformation to a copula, because a copula doesn’t involve causation to start with.

[Travis B.]

I have decided I don't like the use of verbification prefixes for converting adjectives and nouns to predicative verbs, so I changed Rihalle Kaafi so that one can directly use adjectives and nouns in their predicative senses as if they were stative verbs, including being able to receive verbal affixes and agreement clitics, without any kind of verbification affixes or copula. However, verbification prefixes are still used for things like deriving resultative verbs from adjectives and nouns (e.g. 'red' (adj.) zurr > 'redden' (v. pfv.) q'azurr).

[Travis B.]

Several comments:

• Grammaticalisation shifts are generally not reversible. They almost invariably go from the more concrete to the more abstract, not the other way around.
• Beware that Grice’s maxims are not at all universal: they apply well to Western culture, but not so well elsewhere. (Wierzbicka’s Cross-Cultural Pragmatics is a good read.)
• For the same reason I doubt that Grice’s maxims are required to justify grammaticalisation processes. The World Lexicon of Grammaticalisation lists examples of ‘locative copula > exist’ from English, ǁAni, Swahili and Nubi, so it’s not restricted to any one culture.

Thanks for the corrections, I'm a bit of noob, but it's good to know I can simply steal this from English(, Swahili, etc.).

One other thing I should mention: not all of these constructions are precisely the same as the English one, because the other languages here have a dedicated locative verb meaning ‘to be at’. Thus, the Swahili verb -ko means ‘to be at’ with a locative complement, or ‘to exist’ when no complement is specified. In Nubi, the verb fí is transitive meaning ‘to be at’, but in a topic construction it’s existential. And so on.

(It doesn’t even have to be a copular verb: in French the equivalent construction is il y a X, literally meaning ‘it has X there’.)

Anyway, the anticausative meaning more fully aligns with my original thoughts, which had e.g. 'rain' acting more patientive than agentive syntacto-semantically (or however you term it).

That makes no sense to me. The arguments of a copula aren’t agents or patients in the first place — those descriptions don’t fit with their actual semantic roles in a copulative clause. (Which is one reason why, cross-linguistically, copulae often differ from normal verbs in their behaviour, especially when it comes to transitivity.) You can’t apply an anticausative transformation to a copula, because a copula doesn’t involve causation to start with.

It should also be emphasized that it is very common for predicative copulas to differ in other ways from normal verbs, e.g. in many Indo-European languages the predicative copula does not take accusative case arguments (e.g. in German, where sein takes two nominative-case arguments when used with two NP's), and many languages do not even have a predicative copula but rather use verbless clauses in their place.

[bradrn]

It should also be emphasized that it is very common for predicative copulas to differ in other ways from normal verbs, e.g. in many Indo-European languages the predicative copula does not take accusative case arguments (e.g. in German, where sein takes two nominative-case arguments when used with two NP's)

This is precisely the sort of thing I was referring to.

[ophois]

Several comments:

• Grammaticalisation shifts are generally not reversible. They almost invariably go from the more concrete to the more abstract, not the other way around.
• Beware that Grice’s maxims are not at all universal: they apply well to Western culture, but not so well elsewhere. (Wierzbicka’s Cross-Cultural Pragmatics is a good read.)
• For the same reason I doubt that Grice’s maxims are required to justify grammaticalisation processes. The World Lexicon of Grammaticalisation lists examples of ‘locative copula > exist’ from English, ǁAni, Swahili and Nubi, so it’s not restricted to any one culture.

Thanks for the corrections, I'm a bit of noob, but it's good to know I can simply steal this from English(, Swahili, etc.).

One other thing I should mention: not all of these constructions are precisely the same as the English one, because the other languages here have a dedicated locative verb meaning ‘to be at’. Thus, the Swahili verb -ko means ‘to be at’ with a locative complement, or ‘to exist’ when no complement is specified. In Nubi, the verb fí is transitive meaning ‘to be at’, but in a topic construction it’s existential. And so on.

(It doesn’t even have to be a copular verb: in French the equivalent construction is il y a X, literally meaning ‘it has X there’.)

Anyway, the anticausative meaning more fully aligns with my original thoughts, which had e.g. 'rain' acting more patientive than agentive syntacto-semantically (or however you term it).

That makes no sense to me. The arguments of a copula aren’t agents or patients in the first place — those descriptions don’t fit with their actual semantic roles in a copulative clause. (Which is one reason why, cross-linguistically, copulae often differ from normal verbs in their behaviour, especially when it comes to transitivity.) You can’t apply an anticausative transformation to a copula, because a copula doesn’t involve causation to start with.

It's kinda hard to get across what I mean, but in both existentials and anticausatives the subject isn't doing anything at all. It's semantically closer to a patient than an agent. If this was a fluid-S language, the first example, i.e. the one I proposed in my first post, would mark rain as agentive and the other (my preferred system) would mark rain as patientive. It's basically the difference 'rain actively exists' (it chooses to exist, it existing is more relevant than its specific characteristics somehow) and 'rain exists' (the normal kind).

[jal]

I decided that these words with nasal+stop clusters come from a substrate language that allowed these clusters word-finally. Thus the behavior is an indication of the words' foreign origins.

Possible, though typically gross violations of phonology in foreign words lead to adjusting those words to a more native phonology.


JAL

[bradrn]

I decided that these words with nasal+stop clusters come from a substrate language that allowed these clusters word-finally. Thus the behavior is an indication of the words' foreign origins.

Possible, though typically gross violations of phonology in foreign words lead to adjusting those words to a more native phonology.

Either that, or the native phonology itself shifts to accommodate the foreign words.

[jal]

Either that, or the native phonology itself shifts to accommodate the foreign words.

I would only see that happening in case of intensive contact and bilingualism, which would leave a lot more evidence in a language than a few odd stems.


JAL

[Travis B.]

Either that, or the native phonology itself shifts to accommodate the foreign words.

I would only see that happening in case of intensive contact and bilingualism, which would leave a lot more evidence in a language than a few odd stems.

A good example of this occurring in a natlang without intensive bilingualism is the borrowing of nasal vowels from French by German, as I highly doubt that a significant portion of the German population ever spoke French.

[bradrn]

Either that, or the native phonology itself shifts to accommodate the foreign words.

I would only see that happening in case of intensive contact and bilingualism, which would leave a lot more evidence in a language than a few odd stems.

A good example of this occurring in a natlang without intensive bilingualism is the borrowing of nasal vowels from French by German, as I highly doubt that a significant portion of the German population ever spoke French.

I found this hard to believe so I looked this up, and it looks like German only has nasal vowels in the same sense that English does, i.e. not really.

Now getting back to this:

It's kinda hard to get across what I mean, but in both existentials and anticausatives the subject isn't doing anything at all.

What do you mean by ‘not doing anything at all’? I mean, it’s true in that the resulting subjects aren’t agentive, but beyond that they’re certainly not equivalent. I believe that in an anticausative the subject would be a patient, while in an existential it would be… well, honestly I’m not sure. Undergoer? Certainly there’s a difference in that the subject of an existential isn’t affected by the action like a patient is. It’s more like the object of a verb like ‘see’, which is in general completely unaffected in the event.

It's basically the difference 'rain actively exists' (it chooses to exist, it existing is more relevant than its specific characteristics somehow) and 'rain exists' (the normal kind).

I don’t buy this distinction between ‘actively exists’ and ‘(just) exists’. Are there any languages which regularly distinguish these two?

[Travis B.]

I would only see that happening in case of intensive contact and bilingualism, which would leave a lot more evidence in a language than a few odd stems.

A good example of this occurring in a natlang without intensive bilingualism is the borrowing of nasal vowels from French by German, as I highly doubt that a significant portion of the German population ever spoke French.

I found this hard to believe so I looked this up, and it looks like German only has nasal vowels in the same sense that English does, i.e. not really.

To reference a highly reliable source of linguistics information, StG has borrowed nasal vowels from French, but in many cases in actual speech today they have broken into sequences of lax vowels followed by [ŋ] or [n] (compare with modern Polish in this regard). This still represents, though, a greater degree of adaptation to French phonology in loans than English has undergone with more recent French loans, where nasal vowels are normally not borrowed as such but rather are spelling-pronounced when borrowed into English.

[ophois]

I would only see that happening in case of intensive contact and bilingualism, which would leave a lot more evidence in a language than a few odd stems.

A good example of this occurring in a natlang without intensive bilingualism is the borrowing of nasal vowels from French by German, as I highly doubt that a significant portion of the German population ever spoke French.

I found this hard to believe so I looked this up, and it looks like German only has nasal vowels in the same sense that English does, i.e. not really.

Now getting back to this:

It's kinda hard to get across what I mean, but in both existentials and anticausatives the subject isn't doing anything at all.

What do you mean by ‘not doing anything at all’? I mean, it’s true in that the resulting subjects aren’t agentive, but beyond that they’re certainly not equivalent. I believe that in an anticausative the subject would be a patient, while in an existential it would be… well, honestly I’m not sure. Undergoer? Certainly there’s a difference in that the subject of an existential isn’t affected by the action like a patient is. It’s more like the object of a verb like ‘see’, which is in general completely unaffected in the event.

They're both not actively doing anything. If there was syncretism between two of agent, patient, and... undergoer? Theme?, it'd be most natural to fuse the latter two IMO.

It's basically the difference 'rain actively exists' (it chooses to exist, it existing is more relevant than its specific characteristics somehow) and 'rain exists' (the normal kind).

I don’t buy this distinction between ‘actively exists’ and ‘(just) exists’. Are there any languages which regularly distinguish these two?

I sorta confused myself into forgetting my own point here . At least that's proof it wasn't very salient anyway.

[bradrn]

A good example of this occurring in a natlang without intensive bilingualism is the borrowing of nasal vowels from French by German, as I highly doubt that a significant portion of the German population ever spoke French.

I found this hard to believe so I looked this up, and it looks like German only has nasal vowels in the same sense that English does, i.e. not really.

To reference a highly reliable source of linguistics information, StG has borrowed nasal vowels from French, but in many cases in actual speech today they have broken into sequences of lax vowels followed by [ŋ] or [n] (compare with modern Polish in this regard).

Yes, that’s the article I looked at.

This still represents, though, a greater degree of adaptation to French phonology in loans than English has undergone with more recent French loans, where nasal vowels are normally not borrowed as such but rather are spelling-pronounced when borrowed into English.

Perhaps, though don’t some English-speakers use the nasal vowels?

They're both not actively doing anything. If there was syncretism between two of agent, patient, and... undergoer? Theme?, it'd be most natural to fuse the latter two IMO.

‘Not actively doing anything’ is simply a description of every semantic role which isn’t Agent: for instance it also fits Locative perfectly well. Half the point of semantic roles is to distinguish between different ways that participants can ‘not actively do’ things!

(Also, ‘syncretism between semantic roles’ doesn’t really make sense, since the vast majority of languages don’t directly reflect the semantic roles to start with.)