Linguistic Miscellany Thread

[chris_notts]

That's an excellent discussion of topic and focus, thanks!

??As for a pizza, I didn't eat it yesterday

But it's fine to say "As for pizza, I love it."

As I think about it, I'm tempted to call "pizza" here definite, or maybe pseudo-definite. We can have a long conversation about pizza, where it's obviously the topic, and never say "the pizza". We're not talking about one specific entity, but we are talking about one specific classification within the larger category of food.

In support of this, many languages would use the definite article here— e.g. Spanish "Me encanta la pizza."

Yes, I'd say pizza is generic or represents a definite category here, instead of some non-specific or indefinite specific instance of pizzahood.

[bradrn]

And now, a question on a completely different topic (pun not intended): I always assumed that ‘topic’ and ‘focus’ were two words for the same thing, but earlier today I was reading Wikipedia, which mentions that they are different. What are topic and focus really, and how do they differ?

…

Thanks for the detailed reply! I feel I have a much better understanding of this area now. Two questions:

• You say that old topics are usually marked minimally. But what about topic-prominent languages, which always mark the topic?
• Your reply seems to imply that languages either mark topic or focus, but not both. Is this correct? (I ask because I’m making a new conlang with topic-prominent word order and ergative alignment, and I was thinking of making the ergative marking optional to mark focus.)

Another point: referentiality interacts with topic and focus. For obvious reasons, a non-specific, non-generic NP cannot be topical because it makes no sense to say something "about" an entity that the hearer can't identify. A clause like "I eat pizza on Friday night" isn't normally a statement about the unidentifiable pizzas that get eaten in the process. Even worse are non-specifics in irrealis contexts:

I might eat a pizza tomorrow
I didn't eat a pizza yesterday
?? As for a pizza, I might eat it tomorrow
??As for a pizza, I didn't eat it yesterday

Wouldn’t the last two sentences here be examples of a focus rather than a topic? I can easily see them being used in discourse for contrastive emphasis: ‘I will eat a hamburger tomorrow. And as for a pizza, I might eat it tomorrow as well’.

@bradrn: I don't know if it will help, but here's how I've tried to use these kind of typological books in my own grammar: https://www.verduria.org/viewtopic.php?f=3&t=595

Thanks! I don’t think this will help me too much, though: I’m mainly interested in using those books to learn more about linguistics, rather than using them to structure my grammar. (In fact, I feel very strongly that using a template is pretty much the worse way to write a grammar, since it obscures the unity of any language-specific areas which could be used to structure the grammar.)

[akam chinjir]

That was really great, chris_notts. I'm going to highlight/add some things that have tripped me up in the past (and that I hope I now understand well enough!).

Regarding the "pizza" example, my impression is that it's generally true that generics pattern with definites in getting used as topics. But I wonder if it can depend on how they're syntactically encoded? English generics are usually encoded as indefinites, I wonder if there are languages that do that and and mark topics somehow and don't allow indefinitely-marked generics to be topics.

There's a kind of topic that you don't specifically mention but were probably thinking about, ones that sort of set the scene or the context but don't get interpreted as constituents of the subsequent sentences; in English you generally need something like "as for" for these: "As for drinks, I'm a teetotaler."

One thing that's distinctive about this sort of topic is that if you do want it to be interpreted within the subsequent sentence, you generally need a resumptive pronoun: "As for alcohol, I'm against it." (I think this is a pretty robust crosslinguistic pattern, at least in languages that don't allow pro-drop of the relevant arguments.)

My impression is that every language is going to have one or more constructions that can be used with context-setting topics of this sort. It might be far from fully grammaticalised, like maybe you'd just say the equivalent of "if we were talking about X..." (and actually I think it's reasonably common to put conditionals to work in contexts like this), but your language is probably going to have some way to express topics of this sort.

Whereas I'm not sure it's especially common to have a special way (other than pronominalisation) of flagging something as a continuing discourse topic. Chris mentioned passives, there's also simple fronting, maybe with a topic marker of some sort. (I'm thinking of Japanese wa, but actually I don't know enough about how that works.) But even if you've got something like that, I think you're also probably going to have a heavier, less phonologically and syntactically integrated sort of topic marking that can be used for more context-setting topics. And I've somehow picked up the idea that if you've got both sorts of strategies, things like contrastive topics or switch topics might get treated either way, but are unlikely to get an additional topicalisation strategy.

Aside: If anyone knows me to be wrong about this, I'd love to hear about it. I try to keep separate files on things I've learned from extensive reading about many languages, and things that help me organise my thoughts on conlangs, but, well, you know.

I'd also love to hear about it if anyone knows of a language where topics are displaced but not to a position near the beginning of a clause; or where there's an overt topic marker that can be used without movement.

One reason I wonder about that is that it's common to mark focus with an overt particle but no movement, and (at least in SOV languages) focus movement often targets a position immediately before the verb. My impression is that this is something that distinguishes focus from topic, but I can't remember if I've ever seen a general statement to this effect.

I'll add one point about focus. Sometimes when people talk about focus (or emphasis), what they have in mind is an especially interesting or surprising part of a sentence's new information. I'm pretty sure I've seen it claimed that focus of that sort ("informational focus") never triggers movement---it can be flagged by intonation and/or by focus particles, but never by movement. Whereas you can get movement with focus that expresses contrast or exhaustivity, or (I think) when the focused constituent gets linked to a focus-sensitive particle like "only" or "even."

[zompist]

where there's an overt topic marker that can be used without movement.

Quechua. The topic marker -qa can be appended to anything and does not trigger movement.

I'll add one point about focus. Sometimes when people talk about focus (or emphasis), what they have in mind is an especially interesting or surprising part of a sentence's new information. I'm pretty sure I've seen it claimed that focus of that sort ("informational focus") never triggers movement---it can be flagged by intonation and/or by focus particles, but never by movement. Whereas you can get movement with focus that expresses contrast or exhaustivity, or (I think) when the focused constituent gets linked to a focus-sensitive particle like "only" or "even."

Isn't clefting a focus operation? ("It was Jim who refuted Noam.")

[akam chinjir]

where there's an overt topic marker that can be used without movement.

Quechua. The topic marker -qa can be appended to anything and does not trigger movement.

Awesome! Thanks.

I'll add one point about focus. Sometimes when people talk about focus (or emphasis), what they have in mind is an especially interesting or surprising part of a sentence's new information. I'm pretty sure I've seen it claimed that focus of that sort ("informational focus") never triggers movement---it can be flagged by intonation and/or by focus particles, but never by movement. Whereas you can get movement with focus that expresses contrast or exhaustivity, or (I think) when the focused constituent gets linked to a focus-sensitive particle like "only" or "even."

Isn't clefting a focus operation? ("It was Jim who refuted Noam.")

Clefting is focus, for sure, and I definitely should have mentioned it. Is it ever used for purely informational focus?

The question of whether a cleft involves focus movement of any sort is a bit tricky, because relative clauses are often thought to involve movement, at least in many languages, and I'm pretty sure there are analyses according to which "Jim" in "Jim who refuted Noam" has moved from inside the relative clause. (I don't know how they work, though.) One interesting thing is that focus movement tends to pattern with wh movement and relative-clause-forming movement with respect to islands, restrictions on extraction in syntactically ergative languages, and things like weak crossover and parasitic gaps. (Whereas I think there are cases of what looks like topic movement that don't pattern this way; like in some Mayan languages, you get issues with focusing ergative arguments, but never with topicalising them.)

[chris_notts]

Regarding the "pizza" example, my impression is that it's generally true that generics pattern with definites in getting used as topics. But I wonder if it can depend on how they're syntactically encoded? English generics are usually encoded as indefinites, I wonder if there are languages that do that and and mark topics somehow and don't allow indefinitely-marked generics to be topics.

I believe generics can be topics in all languages, but languages differ in whether they're marked like indefinites (English) or definites (Spanish, as zompist said).

There's a kind of topic that you don't specifically mention but were probably thinking about, ones that sort of set the scene or the context but don't get interpreted as constituents of the subsequent sentences; in English you generally need something like "as for" for these: "As for drinks, I'm a teetotaler."

One thing that's distinctive about this sort of topic is that if you do want it to be interpreted within the subsequent sentence, you generally need a resumptive pronoun: "As for alcohol, I'm against it." (I think this is a pretty robust crosslinguistic pattern, at least in languages that don't allow pro-drop of the relevant arguments.)

I've seen these labelled as external vs internal topics. As you say, I think all languages have some way to mark external topics, even if just an intonational break, but not all languages have a marker for internal topics separate from subject promotion.

Whereas I'm not sure it's especially common to have a special way (other than pronominalisation) of flagging something as a continuing discourse topic. Chris mentioned passives, there's also simple fronting, maybe with a topic marker of some sort. (I'm thinking of Japanese wa, but actually I don't know enough about how that works.)

My impression is that wa is there as a topic switcher / activator. The usual expression of an established topic in Japanese is zero. It's almost always, I believe, topic switching that's marked. Another fun way to do it is that some switch reference languages seem to be more sensitive to discourse and topic continuity than actor or subject continuity, but there again it's lack of continuity that normally gets heavier marking.

One reason I wonder about that is that it's common to mark focus with an overt particle but no movement, and (at least in SOV languages) focus movement often targets a position immediately before the verb. My impression is that this is something that distinguishes focus from topic, but I can't remember if I've ever seen a general statement to this effect.

As I mentioned before, my perception is that the emphasis in common between switch topics and contrastive focus often leads to similar marking. In Mayan languages, for example, both are fronted, but there are more subtle morphological and syntactic differences between the two.

I'll add one point about focus. Sometimes when people talk about focus (or emphasis), what they have in mind is an especially interesting or surprising part of a sentence's new information. I'm pretty sure I've seen it claimed that focus of that sort ("informational focus") never triggers movement---it can be flagged by intonation and/or by focus particles, but never by movement. Whereas you can get movement with focus that expresses contrast or exhaustivity, or (I think) when the focused constituent gets linked to a focus-sensitive particle like "only" or "even."

I think that non-contrastive focus is generally not marked explicitly. Topic - comment is more or less the default, so why mark it?

[bradrn]

Looks like my questions got a bit buried in all the discussion, so I’ll repost them in case no-one saw them:

• You [chris_notts] say that old topics are usually marked minimally. But what about topic-prominent languages, which always mark the topic?
• Your reply seems to imply that languages either mark topic or focus, but not both. Is this correct? (I ask because I’m making a new conlang with topic-prominent word order and ergative alignment, and I was thinking of making the ergative marking optional to mark focus.)

(Hopefully reposts like this are OK, but if not I’ll delete the posts.)

[akam chinjir]

I believe generics can be topics in all languages, but languages differ in whether they're marked like indefinites (English) or definites (Spanish, as zompist said).

In languages without definite articles, they can also be unmarked for definiteness.

I've seen these labelled as external vs internal topics. As you say, I think all languages have some way to mark external topics, even if just an intonational break, but not all languages have a marker for internal topics separate from subject promotion.

Yeah, I've seen similar language ("high" topics vs "low" ones).

My impression is that wa is there as a topic switcher / activator. The usual expression of an established topic in Japanese is zero. It's almost always, I believe, topic switching that's marked.

Ah, that sounds reasonable. You can also get double-ga constructions, right?

I think that non-contrastive focus is generally not marked explicitly. Topic - comment is more or less the default, so why mark it?

You can get topic particles and intonation differences to flag especially interesting or surprising new information, though.

• You [chris_notts] say that old topics are usually marked minimally. But what about topic-prominent languages, which always mark the topic?
• Your reply seems to imply that languages either mark topic or focus, but not both. Is this correct? (I ask because I’m making a new conlang with topic-prominent word order and ergative alignment, and I was thinking of making the ergative marking optional to mark focus.)

I'd say that fronting the topic (which can be all that's involved) is pretty minimal.

I didn't see where Chris says that, but certainly you can have both. English has both, for example, if you could "as for" constructions as marking topics and clefts as marking focus. And even if you're only interested in cases where these things involve movement, Mayan languages are an example.

(I was actually just trying to remember if Mayan languages tend to let you front both topic and focus in the same clause. I'm pretty sure I've read that Mayan topic movement targets a position further out than topic movement does, so that if you get both the focus should come before the topic. But I'm not 100% sure of myself.)

[chris_notts]

Looks like my questions got a bit buried in all the discussion, so I’ll repost them in case no-one saw them:

• You [chris_notts] say that old topics are usually marked minimally. But what about topic-prominent languages, which always mark the topic?
• Your reply seems to imply that languages either mark topic or focus, but not both. Is this correct? (I ask because I’m making a new conlang with topic-prominent word order and ergative alignment, and I was thinking of making the ergative marking optional to mark focus.)

(Hopefully reposts like this are OK, but if not I’ll delete the posts.)

Topic prominent language mark topics when they're present. In most I'm aware of, old established topics are not expressed overtly at all. "Topic prominent" is a term invented by a couple of linguists (Li and Thompson) so the category is not, I think, necessarily completely nailed down in a peer reviewed / established terminology kind of way. I have more to say later but I don't have time now.

Languages can mark both topic and some kinds of focus, and I think most or even all have some means of doing so. Specifically, contrastive focus will generally be overtly marked somehow, even if normal information focus is not.

[bradrn]

• You [chris_notts] say that old topics are usually marked minimally. But what about topic-prominent languages, which always mark the topic?
• Your reply seems to imply that languages either mark topic or focus, but not both. Is this correct? (I ask because I’m making a new conlang with topic-prominent word order and ergative alignment, and I was thinking of making the ergative marking optional to mark focus.)

I'd say that fronting the topic (which can be all that's involved) is pretty minimal.

Good point.

I didn't see where Chris says that, but certainly you can have both. \

He doesn’t — I said he ‘seems to imply’ it. I think I got that impression from where he said that most languages pattern as either ‘predicate focus’ (with a topic), ‘argument focus’ (with a focus), or ‘sentence focus’ (no topic or focus).

English has both, for example, if you could "as for" constructions as marking topics and clefts as marking focus. And even if you're only interested in cases where these things involve movement, Mayan languages are an example.

I’m actually really surprised that there’s a language where both topic and focus are both marked by movement — I would have thought that a language which marks both would mark them in different ways.

(As it happens, I’m not particularly interested in cases where they both are marked by movement; I was asking because, as I briefly mentioned earlier, I have a topic-prominent conlang, and I was wondering whether I could mark focus with optional ergative marking.)

Topic prominent language mark topics when they're present. In most I'm aware of, old established topics are not expressed overtly at all.

To clarify, do you mean that most topic-prominent languages are pro-drop? (I suppose that they would be NP-drop, if they can drop any ‘old established topics’.)

"Topic prominent" is a term invented by a couple of linguists (Li and Thompson) so the category is not, I think, necessarily completely nailed down in a peer reviewed / established terminology kind of way. I have more to say later but I don't have time now.

That’s interesting! I had always assumed that ‘topic-prominent’ was a well-established term by now.

Languages can mark both topic and some kinds of focus, and I think most or even all have some means of doing so. Specifically, contrastive focus will generally be overtly marked somehow, even if normal information focus is not.

A question about this: is it possible to have a method of overtly marking contrastive focus, but only on some NPs? And is it possible to have two or more methods of marking contrastive focus? (This is related to the aforementioned topic-prominent conlang with optional ergative marking, where the optional ergative marking only extends across part of the animacy hierarchy.)

[Qwynegold]

What I meant was that S=A or S=O implies that S is being marked similarly to A or O in some way. So either you have ambitransitive verbs where S is variously equated to A or O (e.g. I break the window vs It breaks means that break is an S=O ambitransitive verb), or you have some sort of active-stative alignment where S is variously marked the same as A or O. You have neither: none of your verbs are ambitransitive, and S is always marked the same as O.

Oh, okay. This is only about morphology then.

True — I didn’t notice that. One question about the causatives: is there any particular reason why A-intransitives (maybe you could call them that?) and transitive verbs have a ‘causer’ and ‘causee’, but P-intransitives have an ‘agent’ and ‘patient’?

You could call them causer and causee too. I just wrote like this because my standard average European brain thinks this like a derivational operation more than a grammatical one. Like in 1SG-ERG water-ABS freeze-CAUS I think about it like "I freeze the water" rather than "I make the water freeze". Is it confusing when I use different words for that one type of verb?

[chris_notts]

I’m actually really surprised that there’s a language where both topic and focus are both marked by movement — I would have thought that a language which marks both would mark them in different ways.

(As it happens, I’m not particularly interested in cases where they both are marked by movement; I was asking because, as I briefly mentioned earlier, I have a topic-prominent conlang, and I was wondering whether I could mark focus with optional ergative marking.)

IIRC, many Mayan languages have both (new/contrastive) topic and focus fronting, but when both occur the order is strictly TOPIC FOCUS VERB OTHER. Remember that Mayan languages are mostly verb-initial, so both new topics and foci are marked by movement to pre-verbal position. The difference between the two is marked by the fact that, amongst other things, focus fronting triggers changes to verbal morphology which don't occur when topics are fronted.

You might be interested to know that Tariana has a case system a bit like what you suggest. There is a suffix -n(h)e that applies to focused / non-topical subjects, and a suffix -nuku that applies to unexpectedly topical non-subjects. This case system effectively marks subjects and objects with unexpected pragmatic status only, since subjects are normally topical and objects are more likely to be non-topical.

Many languages have differential object marking, where objects show differences in case marking or verbal agreement depending on factors such as definiteness, specificity, and/or topicality. Differential subject marking is rarer, but does occur, as in Tariana.

To clarify, do you mean that most topic-prominent languages are pro-drop? (I suppose that they would be NP-drop, if they can drop any ‘old established topics’.)

Most of the languages commonly given as examples are, including Japanese and the various East Asian languages people think of. But even a language which isn't will have a most reduced form of reference, whether it's verbal agreement, a clitic pronoun, or just zero. Whatever that most reduced form is will be used with established, predictable topics.

That’s interesting! I had always assumed that ‘topic-prominent’ was a well-established term by now.

Well, maybe I'm wrong, that's just my point of view. I read the Li and Thompson paper a long time ago, but from what I remember topic prominent case felt a bit like an elsewhere case, defined in opposition mainly by its lack of the features in subject prominent languages. It seems to me that there are some things all languages need to do, including:

1. Marking local / clausal topics
2. Tracking interclausal reference (which is strongly related to tracking discourse topics)
3. Marking the roles of arguments in the clause (actor, patient, ...)

Subject prominent languages adopt a solution to this where all of the above are heavily entangled with the subject grammatical relation. The assignment of the subject marks semantic role (normally actor), topicality, and is also often involved in reference tracking via control, switch-reference marking, etc. And because the subject role is marking both semantic and pragmatic roles, some method is needed to mark when these conflict and subject is allocated to an unexpected semantic role. That's where the various grammatical voices like passive and anti-passive come in. Things become a bit messier with ergative languages, where surface case marking and agreement don't align with the typically topical S/A semantic role, but even these may exhibit wide-spread syntactic control phenomena correlated with topicality, linked either to a covert subject role or to the absolutive role.

Topic prominent languages are simply languages which don't entangle pragmatics and semantic roles to the same extent. But it seems to me that there are lots of different ways to do this. If pragmatics isn't marked by grammatical relations then it will be marked by something else, but that something else may be word order, intonation, topic particles, .... Because in these languages GR assignment is not influenced strongly by pragmatics, grammatical voice is not used to reassign GRs to topical referents, and therefore voice is used less and more to eliminate or add roles instead of rearranging the ones that are already there. Control phenomena become a less useful way to track topical referents in discourse, and in particular when coordinating clauses, because you can't rely on the subject always being topical, so referent tracking tends to be more pragmatic in a different sense: context and the choice of referential devices is used for inter-clausal referent tracking instead, and there are few or no grammatical constraints on control of zero/null subjects.

But none of this means that topic prominent languages can't exhibit control, or have any particular grammatical voice, because the category is not a positive one defined by specific positive features being present, it's more a category defined by the absence of an all-powerful subject GR. The lack of such a GR makes many of these things less useful, but languages very commonly have things that they could seemingly happily live without.

A question about this: is it possible to have a method of overtly marking contrastive focus, but only on some NPs? And is it possible to have two or more methods of marking contrastive focus? (This is related to the aforementioned topic-prominent conlang with optional ergative marking, where the optional ergative marking only extends across part of the animacy hierarchy.)

Sure. Tariana marks topic and focus a bit differently on pronouns compared to regular nouns, for example. But I think you need some way of making any argument contrastive, and the structures are probably going to be formally similar even if not identical. I don't know of a language that uses a focus suffix for nouns for a cleft construction for pronouns, for example. Although maybe I just haven't read enough grammars?

[akam chinjir]

IIRC, many Mayan languages have both (new/contrastive) topic and focus fronting, but when both occur the order is strictly TOPIC FOCUS VERB OTHER.

I remembered the Mayan order being the opposite (and said so earlier), but I've gone and checked, and you have it right. I checked Aissen, Topic and Focus in Mayan, which, interestingly, contrasts Tzotzil and Jakaltek, in which only new or contrastive topics can be fronted, with Tz'utujil, in which continuing topics (including pronouns) can as well (with no relevant change in word order).

[chris_notts]

IIRC, many Mayan languages have both (new/contrastive) topic and focus fronting, but when both occur the order is strictly TOPIC FOCUS VERB OTHER.

I remembered the Mayan order being the opposite (and said so earlier), but I've gone and checked, and you have it right. I checked Aissen, Topic and Focus in Mayan, which, interestingly, contrasts Tzotzil and Jakaltek, in which only new or contrastive topics can be fronted, with Tz'utujil, in which continuing topics (including pronouns) can as well (with no relevant change in word order).

Does this correlate with differences in unmarked word order? I don't remember if there are any Mayan languages which are a bit more SVO / less V-initial.

[chris_notts]

He doesn’t — I said he ‘seems to imply’ it. I think I got that impression from where he said that most languages pattern as either ‘predicate focus’ (with a topic), ‘argument focus’ (with a focus), or ‘sentence focus’ (no topic or focus).

Just to clarify a bit here as well, I think your interpretation is a little off. There's a terminology issue here - does a topic have to be a constituent, or not? The issue is that "all of a clause except this bit" may not naturally form a constituent under many approaches, and some people as a consequence distinguish between topics and backgrounded information.

We also have to distinguish degrees of focus: contrastive focus is strong and limited, information focus is weaker and more diffuse. It's the difference between:

What did you do? I went to the swimming pool
Did you eat pizza or a burger? I ate a burger

But regardless of what terminology we use, the three patterns I mentioned have:

PREDICATE FOCUS: one constituent of the clause is topical, the rest is more or less under (generally "weak") information focus
Example: (I)_TOPIC (ate a pizza)_FOCUS

The point here is that there is focal information in predicate focus clauses, but it just may not be overtly marked because this is generally the unmarked case. The topic, if overt, may be marked by word order, a topic particle, or voice (especially if the topic is not the S/A).

ARGUMENT FOCUS: all but one part of the clause is backgrounded information. Non-focal information may be reduced or omitted if the language allows it, since it's assumed to be part of shared knowledge. The non-backgrounded part is under contrastive ("strong") focus
Example: (I ate)_BACKGROUND (a pizza)_FOCUS
Possible shared knowledge / assumption between speaker and listener: speaker ate something, but listener incorrectly thinks it was a burger
Speaker and hearer know: SPEAKER ATE X
Hearer thinks: X = BURGER
Speaker is actually asserting: X = PIZZA

May be marked by word order, clefts, particles, voice (see Mayan langs for an example of voice driven by focus), or emphatic stress on the focused constituent. Clefts in particular highlight the backgrounding of all but one element:

(What I ate)_TOPIC (was a pizza)_FOCUS

SENTENCE FOCUS: the entire clause is under ("weak") information focus, there is no overt topic present
Example: (I ate a pizza)_FOCUS
Possible answer to the question: Did anything happen today?

May be marked by word order, particles, voice. Hard to use a true cleft for this case since there's nothing to be clefted.

[bradrn]

Thanks for the detailed reply! As always, I have a couple of questions/comments:

You might be interested to know that Tariana has a case system a bit like what you suggest. There is a suffix -n(h)e that applies to focused / non-topical subjects, and a suffix -nuku that applies to unexpectedly topical non-subjects. This case system effectively marks subjects and objects with unexpected pragmatic status only, since subjects are normally topical and objects are more likely to be non-topical.

I do find that pretty interesting. I see this as confirming a general principle: topic is unmarked on subjects, and focus is unmarked on objects, since new information (focus) is more likely to be mentioned in S and O and old information (topic) is likely to be tracked in S and A. (Source: Dixon, yet again.)

Many languages have differential object marking, where objects show differences in case marking or verbal agreement depending on factors such as definiteness, specificity, and/or topicality. Differential subject marking is rarer, but does occur, as in Tariana.

I get the impression that differential subject marking is more prominent in ergative languages (as e.g. optional ergative marking), while differential object marking is more prominent in accusative languages. There’s more accusative languages, so differential object marking is more common. Tariana is an active-stative language — a system which could be viewed as a combination of accusative and ergative marking — so it makes sense that it gets both differential subject and differential object marking.

To clarify, do you mean that most topic-prominent languages are pro-drop? (I suppose that they would be NP-drop, if they can drop any ‘old established topics’.)

Most of the languages commonly given as examples are, including Japanese and the various East Asian languages people think of. But even a language which isn't will have a most reduced form of reference, whether it's verbal agreement, a clitic pronoun, or just zero. Whatever that most reduced form is will be used with established, predictable topics.

Well, my conlang doesn’t have verbal agreement. I suppose I’ll have to build some other sort of reduced reference form in then.

(Oh, and a question: how do you tell the difference between verbal agreement and a clitic pronoun? Surely they would be very similar.)

… Things become a bit messier with ergative languages, where surface case marking and agreement don't align with the typically topical S/A semantic role, but even these may exhibit wide-spread syntactic control phenomena correlated with topicality, linked either to a covert subject role or to the absolutive role.

Quick question: what do you mean by a ‘covert subject’?

Topic prominent languages are simply languages which don't entangle pragmatics and semantic roles to the same extent. But it seems to me that there are lots of different ways to do this. If pragmatics isn't marked by grammatical relations then it will be marked by something else, but that something else may be word order, intonation, topic particles, .... Because in these languages GR assignment is not influenced strongly by pragmatics, grammatical voice is not used to reassign GRs to topical referents, and therefore voice is used less and more to eliminate or add roles instead of rearranging the ones that are already there.

Control phenomena become a less useful way to track topical referents in discourse, and in particular when coordinating clauses, because you can't rely on the subject always being topical, so referent tracking tends to be more pragmatic in a different sense: context and the choice of referential devices is used for inter-clausal referent tracking instead, and there are few or no grammatical constraints on control of zero/null subjects.

But none of this means that topic prominent languages can't exhibit control, or have any particular grammatical voice, because the category is not a positive one defined by specific positive features being present, it's more a category defined by the absence of an all-powerful subject GR. The lack of such a GR makes many of these things less useful, but languages very commonly have things that they could seemingly happily live without.

I’m not sure I understand what you’re saying here. Are you saying that topic-prominent languages less often have argument-rearranging constructions like passive and antipassive, but are rich in valency-changing constructions like causative and applicative? Or are you saying that topic-prominent languages can still have all these constructions, but don’t usually use them purely for argument rearrangement in the way that most non-topic-prominent languages do?

Also, once you bring up control: I always thought that in a topic-prominent language, the topic could be used as a pivot for control — that is, when you coordinate two clauses, it is the topics that get equated (rather than, say, S/A being equated). (So for example in imaginary topic-prominent English, ‘the man(topic) saw the woman and ran’ would have the man running, but ‘the woman(topic) the man saw and ran’ would have the woman running.) Is this attested? And if so, how common is it?

A question about this: is it possible to have a method of overtly marking contrastive focus, but only on some NPs? And is it possible to have two or more methods of marking contrastive focus? (This is related to the aforementioned topic-prominent conlang with optional ergative marking, where the optional ergative marking only extends across part of the animacy hierarchy.)

Sure. Tariana marks topic and focus a bit differently on pronouns compared to regular nouns, for example. But I think you need some way of making any argument contrastive, and the structures are probably going to be formally similar even if not identical. I don't know of a language that uses a focus suffix for nouns for a cleft construction for pronouns, for example. Although maybe I just haven't read enough grammars?

Well, I was actually planning to use two very different constructions for coordination, so it’s useful to know that won’t work. I’ll have to think of another way of doing this then.

We also have to distinguish degrees of focus: contrastive focus is strong and limited, information focus is weaker and more diffuse. It's the difference between:

What did you do? I went to the swimming pool
Did you eat pizza or a burger? I ate a burger

To make sure I understand: the first example here is informational focus while the second is contrastive focus, right?

But regardless of what terminology we use, the three patterns I mentioned have:

PREDICATE FOCUS: one constituent of the clause is topical, the rest is more or less under (generally "weak") information focus
Example: (I)_TOPIC (ate a pizza)_FOCUS

The point here is that there is focal information in predicate focus clauses, but it just may not be overtly marked because this is generally the unmarked case. The topic, if overt, may be marked by word order, a topic particle, or voice (especially if the topic is not the S/A).

ARGUMENT FOCUS: all but one part of the clause is backgrounded information. Non-focal information may be reduced or omitted if the language allows it, since it's assumed to be part of shared knowledge. The non-backgrounded part is under contrastive ("strong") focus
Example: (I ate)_BACKGROUND (a pizza)_FOCUS
Possible shared knowledge / assumption between speaker and listener: speaker ate something, but listener incorrectly thinks it was a burger
Speaker and hearer know: SPEAKER ATE X
Hearer thinks: X = BURGER
Speaker is actually asserting: X = PIZZA

May be marked by word order, clefts, particles, voice (see Mayan langs for an example of voice driven by focus), or emphatic stress on the focused constituent. Clefts in particular highlight the backgrounding of all but one element:

(What I ate)_TOPIC (was a pizza)_FOCUS

SENTENCE FOCUS: the entire clause is under ("weak") information focus, there is no overt topic present
Example: (I ate a pizza)_FOCUS
Possible answer to the question: Did anything happen today?

May be marked by word order, particles, voice. Hard to use a true cleft for this case since there's nothing to be clefted.

A clarification here: I assumed earlier that each of these patterns was present over a whole language (i.e. much like an ergative alignment may be found on all clauses, similarly a language might always consistently mark focus rather than topic and so be a language with argument focus), but you seem to imply here that these patterns are properties of individual clauses and sentences rather than a property of a language as a whole. Which interpretation is correct?

[akam chinjir]

Does this correlate with differences in unmarked word order? I don't remember if there are any Mayan languages which are a bit more SVO / less V-initial.

Not sure if it'd be strictly unmarked, though presumably there are a lot more preverbal subjects if continuing topics are also regularly fronted, and I'm not sure that a definiteness requirement (say) on preverbal subjects makes that a marked position.

That reminds me of something that I read recently that surprised me a bit, the idea that some indefinites can be topical. In a context something like this: I've just mentioned three people, I then say "One of them did X, another did Y, and the third did Z"; the idea is that it would be possible to treat the subjects in those sentences as topics. They're indefinite but from a particular contextually salient group (i.e. they're partitive), they're a sort of contrastive topic, and I think maybe it's important that the listing is exhaustive.

I get the impression that differential subject marking is more prominent in ergative languages (as e.g. optional ergative marking), while differential object marking is more prominent in accusative languages.

Except that differential subject marking in ergative languages tends to take a form like this: the subject gets ergative case marking only if the object is sufficiently animate or definite or specific or whatever. That's to say, the case-marking on the subject is conditioned by the features of the object. (And in a tripartite language like Nez Perce, differential marking takes a form where if the object is sufficiently definite or whatever, the subject is ergative and the object accusative; but when the object is not sufficiently definite, neither argument gets marked for case.) DSM conditioned by, say, the definiteness of the subject is a lot rarer.

[bradrn]

I get the impression that differential subject marking is more prominent in ergative languages (as e.g. optional ergative marking), while differential object marking is more prominent in accusative languages.

Except that differential subject marking in ergative languages tends to take a form like this: the subject gets ergative case marking only if the object is sufficiently animate or definite or specific or whatever. That's to say, the case-marking on the subject is conditioned by the features of the object. (And in a tripartite language like Nez Perce, differential marking takes a form where if the object is sufficiently definite or whatever, the subject is ergative and the object accusative; but when the object is not sufficiently definite, neither argument gets marked for case.) DSM conditioned by, say, the definiteness of the subject is a lot rarer.

You did mention this before, so I am aware of it. But I’m surprised by your statement that ‘DSM conditioned by … the subject is a lot rarer’: from what I’ve seen, this sort of DSM is just as common, if not more common. To clarify, are you including optional ergative marking (where it is common for the presence of the ergative marker to be related to animacy or focus rather than the animacy of O), or are you only talking about cases where there are multiple ergative markers? I consider DSM to include both.

[akam chinjir]

You did mention this before, so I am aware of it. But I’m surprised by your statement that ‘DSM conditioned by … the subject is a lot rarer’: from what I’ve seen, this sort of DSM is just as common, if not more common. To clarify, are you including optional ergative marking (where it is common for the presence of the ergative marker to be related to animacy or focus rather than the animacy of O), or are you only talking about cases where there are multiple ergative markers? I consider DSM to include both.

I wasn't thinking of ergativity splits conditioned by, say, contrasts between pronominal and NP arguments, at least. If by focus, you mean languages in which the subject doesn't get ergative case when focused, I also wasn't thinking of that. I guess these all seem like quite different phenomena to me, and I'm not used to seeing them called DSM, though that could easily just be gaps in my reading.

Tangentially, I'd be interested to know of a case where focus prevents ergative case-marking---the case I'm relatively familiar with is Mayan languages, but that's agreement rather than case, and antiagreement like that has nothing in particular to do with ergativity.

[bradrn]

You did mention this before, so I am aware of it. But I’m surprised by your statement that ‘DSM conditioned by … the subject is a lot rarer’: from what I’ve seen, this sort of DSM is just as common, if not more common. To clarify, are you including optional ergative marking (where it is common for the presence of the ergative marker to be related to animacy or focus rather than the animacy of O), or are you only talking about cases where there are multiple ergative markers? I consider DSM to include both.

I wasn't thinking of ergativity splits conditioned by, say, contrasts between pronominal and NP arguments, at least. … I guess these all seem like quite different phenomena to me, and I'm not used to seeing them called DSM, though that could easily just be gaps in my reading.

I had assumed that differential subject marking was a general category of phenomena in which the subject can be marked in multiple different ways, encompassing both optional ergative marking and the presence of multiple ergative markers. Is this definition of DSM incorrect?

If by focus, you mean languages in which the subject doesn't get ergative case when focused, I also wasn't thinking of that.

No; rather I was thinking of languages like Lhasa Tibetan, in which the presence of ergative case indicates contrastive focus.