Beyond copulas: 'adverbial' meaning+be?

[Daniel]

Are there any languages that have a wide variety of copula-like elements that describe certain relationships with the complements?

For example, we might gloss some in English along these lines:

I be.very happy.
I be.just.a.little happy.
I be.not happy.
I be.best student.
It be.mostly red.
I be.probably helpful.
etc.

Of course these might not be considered canonical "copulas" with those meanings, but they still seem logically similar.

I suppose the closest examples I can think of are words like "become" in English or "estar" in Spanish, with slight nuances.

[freknu]

Actually, become is not exactly a copula as such. It comes from *bi- (locative prefix) and *kwemaną "to come". Cf. beside, below, beset, belong, begin, etc.

[Daniel]

The term "copula" is somewhat vague. In a restricted sense it basically means an equals sign (a linker between subject and property). But I mean it more semantically.

We can see 'become' as an inceptive aspect copula:
I was happy => I was happy at a particular time.
I became happy => I was happy after a time when I was not happy-- I began being happy.

So "become" is like "begin to be". It certainly has "be" in its meaning in some sense.

What I'm curious about is why we don't find too many of these verbs cross-linguistically.

The etymology is also interesting in how they might develop though.

[freknu]

Then you might as well include verbs like grow, begin, lay, etc., which are hardly rare cross-linguistically. Or is this not actually about "copulas", but auxiliary verbs?

[MalFet]

To echo freknu, I'm not sure the question here is coherently formed yet.

On the one hand, you're using a fairly minimalist notion of copula ("equals sign"), but then you seem to be looking for examples with more elaborate semantics (to include become, for example). In this latter sense, aren't all predicates "copulas" of sorts?

[Daniel]

Then you might as well include verbs like grow, begin, lay, etc., which are hardly rare cross-linguistically. Or is this not actually about "copulas", but auxiliary verbs?

"Grow" seems like "become" to me (perhaps with a slightly slower transition). So that is an interesting example. But the others? They don't seem to take the same complements as copulas. (*I begin happy, *I lay happy.)
There are of course depictives like "arrive tired", etc, but those aren't arguments, just additional descriptions.

To echo freknu, I'm not sure the question here is coherently formed yet.

See the examples in my first post.

On the one hand, you're using a fairly minimalist notion of copula ("equals sign"), but then you seem to be looking for examples with more elaborate semantics (to include become, for example). In this latter sense, aren't all predicates "copulas" of sorts?

Possibly, but a verb like "eat" doesn't equate anything. A verb like "become" does. In set theoretic terms, copulas seem to assign membership to the set represented by their argument to their subject. "I am happy" means that I am among those entities in the set "happy things". The same applies to "a teacher" or "in class".


This isn't a well-formulated research question yet because it's just a thought that I had. To explain the context, I heard a line of dialog in French [TV show] that didn't seem to be well-formed to me. It was a child mispronouncing and whispering, so maybe the copula was elided. But what I heard was "Je très froid" or 'I very cold'. That speech error(?) inspired the question-- I just wondered why I have never come across a copula in a language that means "to be very X" or "to be only X" (and see the list above).
While I have no idea if a language would really benefit from having such forms, they don't seem especially implausible semantically or syntactically. Languages do lots of weird things, so why not that?

To rephrase the question more concretely, then, why do no languages have a verb like "to be very X"?
(If there is a language that has something like that, that's fine too, but I haven't seen one, so I'll operate on the assumption that we won't find one just because the implications might be interesting, not that I would be surprised to find one if I knew where to look.)

[freknu]

A verb that would also have an explicit adjectival/adverbial degree? I don't know about usefulness, but it sure would be extremely restricted.

"I sit sad" (ignore grammaticallity) is not action-neutral, it tells you what I am doing and it tells you my state. "I am sad" is action-neutral, even though there is a verb there that tells you what I am doing (being), but it has become a sort of neutral default state, and thus only the part that tells you my state is relevant.

You could say "I sad", but there seems to be a preference to having an "empty" or "neutral" auxiliary verb (or copula), often from "to exist" or "to stand". "Become" is not neutral, it is explicitly describing an action, you cannot say "I sad" and still retain the tense of "(in future) I sad".

Without the auxiliary verb, the adjective would have to take on tense to show the state in temporal relation, so the reason there is a neutral auxiliary verb is probably to allow temporal constructs.

Nominal-adjectival merger seems to be fairly common cross-linguistically, but verbal-adjectival is not something I've ever heard of.

You are essentially asking adjectives (or adverbs) to carry tense.

In a language with classical word classes it's silly, and in a language without classical word classes it's irrelevant :/

[Daniel]

You could say "I sad", but there seems to be a preference to having an "empty" or "neutral" auxiliary verb (or copula), often from "to exist" or "to stand". "Become" is not neutral, it is explicitly describing an action, you cannot say "I sad" and still retain the tense of "(in future) I sad".

Without the auxiliary verb, the adjective would have to take on tense to show the state in temporal relation, so the reason there is a neutral auxiliary verb is probably to allow temporal constructs.

Nominal-adjectival merger seems to be fairly common cross-linguistically, but verbal-adjectival is not something I've ever heard of.

That's actually very common outside of Europe. Japanese, for example (just one of the more well known languages like this), has a very fuzzy distinction between adjectives and verbs. Some languages have no such distinction at all, unless you want to claim that they theoretically do on semantic grounds.

You are essentially asking adjectives (or adverbs) to carry tense.

In a language with classical word classes it's silly, and in a language without classical word classes it's irrelevant :/

There's nominal tense in some languages, such as Guaraní (South America). Adjectival tense essentially merges adjectives with verbs, though, so it's not even that different.


To put this another way, it seems that most languages have a basic copula that fits all situations, or perhaps have no copula (a "null copula" or however you want to call that), and that doesn't vary paradigmatically depending on the meaning in question. But what if it did? Ser and estar in Spanish are a pretty close example. I don't know that "estar"  is technically a copula, but it's something very close to one and both translate "be" in English. What if there were a language with 12 verbs that translate English "be" all with different nuance, perhaps covering some of what we use adverbs for. It just doesn't seem so unexpected that it would never be found. Or perhaps if it isn't, there's something basic/"universal" about the copula?

[freknu]

Oh, so you're just questioning the number of verbs in other languages that correspond to the copula in English?

[MalFet]

On the one hand, you're using a fairly minimalist notion of copula ("equals sign"), but then you seem to be looking for examples with more elaborate semantics (to include become, for example). In this latter sense, aren't all predicates "copulas" of sorts?

Possibly, but a verb like "eat" doesn't equate anything. A verb like "become" does. In set theoretic terms, copulas seem to assign membership to the set represented by their argument to their subject. "I am happy" means that I am among those entities in the set "happy things". The same applies to "a teacher" or "in class".

By most formal representations, that's how all predication works, though: "I eat cookies" is true iff "I" am in the set of things that "eat cookies". The issue is not as simple as equating things or locating things in sets. By any coherent account, a copula is something much more specific.

And, when we actually begin with that specificity, I think your question unravels a little bit. Semantically, most languages (including English) use a wide range of words to link subjects with non-finite predicates of some sort or another. Most of these would rarely translate back into English as "be", but that's largely because the word be is for all intents and purposes *defined* by its lack of more specific semantics.

If you're just looking for words that mean be+qualification, we do this all the time in English:
This apple exudes red.
John seems humble.
Tim passes muster as a teacher.
That haunted house exemplifies scary.

If you want to argue that these aren't copulae (and there are plenty of reasons that you might), you're going to need a much more explicit definition of copula than you presently have.

[Daniel]

Oh, so you're just questioning the number of verbs in other languages that correspond to the copula in English?

More or less, yes. Or, to rephrase that: why there aren't languages that translate English 'be' with more nuance.

By most formal representations, that's how all predication works, though: "I eat cookies" is true iff "I" am in the set of things that "eat cookies". The issue is not as simple as equating things or locating things in sets. By any coherent account, a copula is something much more specific.

There's a a major difference: "eat" determines a set itself (eating things), while "be" does not ("being things"?). "Be" is without any real lexical meaning, just a grammatical function. "Eat" has real semantic content. "Eat" describes an action or a situation. "Be" does not. "Be very" does not either-- it just modifies some existing set to pick a subset of whatever that set may be.

If you're just looking for words that mean be+qualification, we do this all the time in English:
This apple exudes red.
John seems humble.
Tim passes muster as a teacher.
That haunted house exemplifies scary.

Ok, these are relevant examples. And I think we should try to determine where, if anywhere, a line should be drawn.

To me, "exudes" and "passes muster as" seem too infrequent to be grammaticalized as paradigmatically related to the copula. They seem stylistic and probably have semantic content, though it's abstract (in contrast to "eat" etc.).

"Seem", however, is a great example: I don't feel like this word has any particular lexical meaning-- rather, it is probably a modal copula, describing epistemic possibility (modality) plus 'be'. I would draw a distinction between "seems to be X" and "seems X", where the former is just a modifier and the latter is actually taking the place of the copula, assuming we don't see that as some sort of transparent ellipsis.

"Exemplifies" is intriguing-- it seems like as good a candidate as any for a set theoretic semantic representation, though in those terms it seems like just an equivalent to 'be', rather than in complementary distribution with it. I think there are two readings, one that is just 'be' in a longer word, and another where it means something like 'is the best example of', which would be along the lines of what I'm thinking, but I don't know that it's so established in English to really be considered grammaticalized in that sense. You wouldn't find translators often using "exemplifies" rather than "be", but you would probably find translators often using "seem" rather than "be" if translating from a language with markers of evidentiality.

Most of these would rarely translate back into English as "be", but that's largely because the word be is for all intents and purposes *defined* by its lack of more specific semantics.

As I said, I see a distinction between "be" and "eat", where the latter has true lexical meaning and the former serves a grammatical function, looking for a set in its argument. Likewise "be very" or "seem" would still take the core meaning from their arguments rather than the lexical entry. Compare:
*The things that be. [existential reading excluded]
*The things that are very.
*The things that seem.
The things that eat.
Or:
*Be-ers, *Very be-ers / *Extreme be-ers.
*Seemers
Eaters

So while "be" can take an argument it is not a transitive verb. That is because its argument is the only semantic content. This is still true for "seem" even though it has some (modificational) semantic content. But compare:
Exuders / Exuding things
Those that pass muster.
Exemplifiers / Exemplars

So we're left with [at least] three verbs in English that appear special in some way: be, seem, become
(*Those that become, *becomers)

The only problem with this is that 'seem' (but not 'become'?) can't take all of the kinds of arguments that 'be' does, so it can't be in a true paradigmatic relationship:
He seems happy.
??He seems inside.
*He seems (a) teacher. (?He seems a good man.)

While specific grammatical form isn't proof of anything, one further test is checking what case these verbs require for their arguments, at least on average cross-linguistically. 'be' often takes nominative, while most verbs take accusative. I think that in general verbs like 'become' and 'seem' would also take nominative arguments.

I've done a little research on the 'null copula' in Arabic (just in the present tense). One obvious question is whether it's just not pronounced. Regardless of the analysis you end up with, there is an intriguing property of the Arabic copula: when it is produced (for example, in the past tense), it takes accusative arguments:
He was teacher.ACC 'He was a teacher'
*He was teacher.NOM
This is in contrast to present tense with:
He teacher.NOM 'He is a teacher.'
*He teacher.ACC

This leads to an analysis where we don't consider the present tense to be an instance of a phonologically missing form of 'kaana' (glossed "to be" in dictionaries, etc.), as used in the past tense. Instead, it seems that 'kaana' may not even be a copula, just a lexical verb that establishes tense. Perhaps something like "It was the case that" or some meaning that can't be translated into English.


As for exactly where to draw the line between "copula" or "copula like verb" and "regular verbs" I still don't know. But "seem" is a good example, I think.

The open question: do we find any languages with a clear verb that is used when the speaker wants to emphasize how extremely a property applies to the subject? Is there a language with 'to be very' as a single lexical predicate that is in complementary distribution to 'to be'?

[MalFet]

By most formal representations, that's how all predication works, though: "I eat cookies" is true iff "I" am in the set of things that "eat cookies". The issue is not as simple as equating things or locating things in sets. By any coherent account, a copula is something much more specific.

There's a a major difference: "eat" determines a set itself (eating things), while "be" does not ("being things"?).

This is an extremely unorthodox use of set-based semantics. If you're drawing on a particular model here, you'll have to point it out to me because it is one I'm not familiar with. Within the mainstream Frege->Russell->Montague->Kripke paradigm, however, your account here is deeply incorrect.

"Be" is without any real lexical meaning, just a grammatical function. "Eat" has real semantic content. "Eat" describes an action or a situation. "Be" does not. "Be very" does not either-- it just modifies some existing set to pick a subset of whatever that set may be.

I truly have no idea what you mean by "real semantic content". That's not a notion I've ever seen expressed in any framework of semantics, whether formal, cognitive, or sociolinguistic. Function words have semantics above all else. The obsession that formal semantics has with function words is a common criticism. When you say "content", do you mean "reference"?

Most of these would rarely translate back into English as "be", but that's largely because the word be is for all intents and purposes *defined* by its lack of more specific semantics.

As I said, I see a distinction between "be" and "eat", where the latter has true lexical meaning and the former serves a grammatical function, looking for a set in its argument. Likewise "be very" or "seem" would still take the core meaning from their arguments rather than the lexical entry. Compare:
*The things that be. [existential reading excluded]
*The things that are very.
*The things that seem.
The things that eat.

Well, it'd be "The things that are", not "The things that be", since English still requires subject-verb agreement, and that's a perfectly valid reference. As for the second sentence, the word "very" of course does not modify verbs, so I'm not really sure what you're trying to demonstrate here other than that some combinations of words don't work in English. "The things that seem" is not particularly worse than "The things that free" (such as Moses, who freed his people from Egypt).

The open question: do we find any languages with a clear verb that is used when the speaker wants to emphasize how extremely a property applies to the subject? Is there a language with 'to be very' as a single lexical predicate that is in complementary distribution to 'to be'?

I don't follow how something could be a complete predicate and a copula at the same time, but assuming you mean "single lexical item" instead of "single lexical predicate"...Sure. Off the top of my head, even the vanilla copula "asti" in Sanskrit freely takes adverbs, and you even have words like "ati" that can occupy asti's position and mean "to be very". Several of the signed languages I know do this kind of thing a lot, too.

[Daniel]

This is an extremely unorthodox use of set-based semantics. If you're drawing on a particular model here, you'll have to point it out to me because it is one I'm not familiar with. Within the mainstream Frege->Russell->Montague->Kripke paradigm, however, your account here is deeply incorrect.

I'm not intending the question in particularly formal terms, but for clarity I'll try to rephrase. I suppose what I'm saying is that the set is not restricted in any way by "be". Every entity in the universe can be predicated with 'be' in some way:
X is real.
X is fake.
X is X.
etc.

Not every entity in the universe can be the subject of eat.

So the set determined by 'be' (and 'be' alone) is the set of all entities in the universe.
The set determined by 'eat' is a subset of that set, as is the case with probably all other verbs.

If we modify 'be' in a broad way, it doesn't change that relationship in an especially obvious way (though the tautology may no longer hold).

I truly have no idea what you mean by "real semantic content". That's not a notion I've ever seen expressed in any framework of semantics, whether formal, cognitive, or sociolinguistic.

As I said, I didn't mean that formally within any theory. I meant it intuitively, though perhaps that doesn't communicate it clearly for you.

To me, there is a very different amount of information available about the subjects in these two clauses:
"X is very Y"
"X eats Y"

There's a categorical difference: in the second sentence we know that X is capable of eating and is currently eating something. In the first all we know is that some property of X (the property Y) is very much true. The set of possible Xs is much narrower (in, I think, a categorical way) for the second sentence compared to the first.

Function words have semantics above all else. The obsession that formal semantics has with function words is a common criticism. When you say "content", do you mean "reference"?

Yes, reference is a place to start. "Eat" refers to "eaters", while it seems odd to say that "be" refers to "be-ers". (X is fake does not entail that X exists or has any properties or anything at all really.)

Well, it'd be "The things that are", not "The things that be", since English still requires subject-verb agreement, ...

Oops. Typo-- analyzing rather than relying on my (actual) ability to speak English :p

Did you mean this?--

"The things that are", ... and that's a perfectly valid reference.

That is only a valid sentence in English in the existential sense, where I would consider it no longer to be a copula. "To be or not to be" is not a question of whether one does or does not have any properties but whether one is alive or exists in the real universe. (Existence and copulas are not the same, as shown by, among other things, grammatical patterns in say Quechua where the verb 'to be' is omitted in 3SG.PRES unless it means "exists".)

As for the second sentence, the word "very" of course does not modify verbs, so I'm not really sure what you're trying to demonstrate here other than that some combinations of words don't work in English.

I believe that "very" modifies "be [X]", so you are implying that I'm talking about something that is not properly compositional (not isomorphic to assumed compositional semantic structures), which is a valid point. That might get toward the "why" question of why we don't find "be very" re-lexicalized as a single word too often.

"The things that seem" is not particularly worse than "The things that free" (such as Moses, who freed his people from Egypt).

Really?? I don't get that intuition at all.
While "being something/one that frees" seems to give me relevant information about an individual, "being something/one that seems" is entirely vacuous. It doesn't narrow down any potential set of entities at all-- I can't imagine a world in which "seemers" could be identified compared to "non-seemers", while certainly "freers" can be identified in contrast to "non-freers".
(I suppose there's some bizarre level of abstraction where, in the right context, anything/one can be anything, but if we rely on the common assumption that "red" refers to the set of "red things" and that it is some subset of entities, I do think my point holds.)

I don't follow how something could be a complete predicate and a copula at the same time, but assuming you mean "single lexical item" instead of "single lexical predicate"...Sure.

Hmm... you're right about that. I was trying to avoid the word "verb" because it's arguable that "copulas" are not "verbs" (this terminology is a mess), but "predicate" isn't better anyway. I suppose I'm looking for the word for "that which is located as a head on the assumed C-T-V spine of a syntactic tree"-- so "verbs and friends". Should we call it an auxiliary? That doesn't seem right either. But maybe it's closer. Anyway, "item" is just fine.

-----

Off the top of my head, even the vanilla copula "asti" in Sanskrit freely takes adverbs, and you even have words like "ati" that can occupy asti's position and mean "to be very". Several of the signed languages I know do this kind of thing a lot, too.

Ok, so confusing technical discussion aside, this seems to be getting at exactly what I was wondering. Can you explain a bit more or give some examples via glosses?

These languages have, I assume [and ignoring word order-- probably SOV?], something like:
Subj BE [Adj/N/etc]

And you're saying they can instead have:
Subj VERY [Adj/N/etc]

So would that "VERY" be more like the actual lexical item 'very', as in "I very happy", or a grammaticalized verb-like element, "I be.very happy"? In other words, is this simple ellipsis, or is there a grammaticalization process going on?

[MalFet]

This is an extremely unorthodox use of set-based semantics. If you're drawing on a particular model here, you'll have to point it out to me because it is one I'm not familiar with. Within the mainstream Frege->Russell->Montague->Kripke paradigm, however, your account here is deeply incorrect.

I'm not intending the question in particularly formal terms, but for clarity I'll try to rephrase. I suppose what I'm saying is that the set is not restricted in any way by "be". Every entity in the universe can be predicated with 'be' in some way:
X is real.
X is fake.
X is X.
etc.

Not every entity in the universe can be the subject of eat.

So the set determined by 'be' (and 'be' alone) is the set of all entities in the universe.
The set determined by 'eat' is a subset of that set, as is the case with probably all other verbs.

If we modify 'be' in a broad way, it doesn't change that relationship in an especially obvious way (though the tautology may no longer hold).

This is just...not how it works.

I apologize if it seems I'm just being pedantic, but set-based approaches to semantics are incoherent if the terminology isn't carefully defined and applied. What "can be" a subject of what, what is or isn't "a subset" of what, which parts of a predicate constitute the set, "how much" semantic content a predicate has...these are deeply problematic misconstruals of formal semantic analysis. I understand that you're trying to remain theory-neutral, but you can't have your cake and eat it too. You can't hope to have meaningful cross-linguistic comparison of semantic equivalence and then wave away the problem of equivalence as too technical to be interesting. If you're just going on your intuitions, what you end up with is at best English-centric and at worst a quagmire of contradictions.

For example, I still cannot for the life of me understand what you intend by the word "copula". On the one hand, you seem to start with a fairly standard definition: a word used to "link" a subject with a non-finite subject complement. Fine enough. As I understand it, you also want the term to be limited to syntactic verbs. That eliminates a few things that have conventionally been called copulas, but that's also fine. You start running into problems, though, when you try to eliminate "exudes" for being "too infrequent to be grammaticalized", or "exemplifies" because translators wouldn't find it a ready substitute for "be". If grammaticalization in a lexical-syntactic sense is a necessary basis of a copula, I have no idea how to interpret your initial question.

A word is grammaticalized (in formal terms) precisely to the extent that it loses its referential qualities to become a non-referential logical operator. To begin with this definition and then to ask why there aren't any copulas with relatively more narrow referential semantics is tautological. It's like asking why an entity can't be both A and ~A. If (against convention) you want to limit copular functions to low-level function words like this, of course you aren't going to find copulas with elaborate semantics. That's not a descriptive fact about the world; it's an inevitable consequence of our very particular way of framing the ontology of formal logic.

Off the top of my head, even the vanilla copula "asti" in Sanskrit freely takes adverbs, and you even have words like "ati" that can occupy asti's position and mean "to be very". Several of the signed languages I know do this kind of thing a lot, too.

Ok, so confusing technical discussion aside, this seems to be getting at exactly what I was wondering. Can you explain a bit more or give some examples via glosses?

These languages have, I assume [and ignoring word order-- probably SOV?], something like:
Subj BE [Adj/N/etc]

And you're saying they can instead have:
Subj VERY [Adj/N/etc]

So would that "VERY" be more like the actual lexical item 'very', as in "I very happy", or a grammaticalized verb-like element, "I be.very happy"? In other words, is this simple ellipsis, or is there a grammaticalization process going on?

As I've said, I don't understand the basis of this distinction you're making, so I can't answer your question. But, in Sanskrit you can use all sorts of verb-like copulas, such as "atibhu":

(sandhi removed for simplicity's sake)
1)    ramaḥ      biraḥ      as-ti
      Ram       brave      be-3s
      "Ram is brave"

2)    ramaḥ      biraḥ      atibhava-ti
      Ram       brave      very.be-3s
      "Ram is very brave"
      
Both sentences' third words are verbs. Whether they fit your intuitions about what makes a copula, I can't guess, but both are functional copulas by any conventional definition. You can do similar things in Nepali and Hindi (and I would assume many other Indic languages).

[Daniel]

For example, I still cannot for the life of me understand what you intend by the word "copula".

I'm not trying to make it complicated: as I said in the title of this thread, be+[adverbial].
So terminology aside, let me rephrase my question as simply asking whether there are languages out there that have words equivalent to 'be' except that they have more nuance.

It's like asking why an entity can't be both A and ~A. If (against convention) you want to limit copular functions to low-level function words like this, of course you aren't going to find copulas with elaborate semantics.

Correct. And the meaning of "copula" is vague (though I mean it generally in its etymological sense, "connecting word"). I'm just using it as a term for the English lexical item BE and similar items in other languages.


If you're unsure of my terminology, then how would you classify the words "be", "become" and "seem" [in their linking sense]? To me they form a very different kind of group than, say, "free", "eat" and "swim".


----
More importantly, the Indic case:

That's exactly what I was wondering about. Is this because of a systematic underspecification of the adverb/verb[/BE] classes or is this a case of lexicalization/grammaticalization from one class to another?
You're calling them "verbs", so I'd imagine it's the latter.

So in short, thanks! That's exactly what I was wondering about.

Any idea of there's a description/typology of those verbs? It would be interesting to take a look through them.


(Note: this isn't for research, just a curiosity based on (mis)hearing a sentence on TV.)