 We are very pleased and very happy to have our guest speaker this afternoon, Kaya Jero. Kaya is a semanticist. He works on learning the semantics. He also works on syntax and pragmatics. He is a research expert. He is a language expert. He has worked extensively on Kinga Wanda and has conducted food work in Wanda several times. Kinga Wanda is a bigger mantra language and there is quite some work already which has been done. There is no Chemienghi in the States. There is no Belgian work. But then Kaya went back and revisited some of the classic topics in the office in the sense of mantra kinga Wanda. And actually found that there is much more variation in that, much more undetected detail. And so that has brought Bantu scholarship to a higher level than the classic situation where you want to truly work in the language. And he thinks that's it. And it really shows very well that revisiting these things that we think we know quite a lot about actually turn out to be much more complex than they are. His PhD is from the University of Austin in Texas. He then had a small poster and then he almost came to saw us. When we got funding for Kaya to join us yet, but then he got headhunted, which is not our language. It's not the sort of field where that happens. So he then got headhunted by the University of Essex, where he is now a lecturer in linguistics. And through that and also the movement of Hanna, who also comes from here to Essex, Essex is now developing into the center of Bantu linguistics in the UK. I'm not expecting that, but still, sorry. So that's very exciting, but it's exciting of course for us in the UK as a community. So we are very pleased that Kaya is in the UK at Essex. And we're very pleased that he's here this afternoon. He's going to talk about graph media and various exchange morphology insights from Bantu. Thank you very much. Thank you for coming. So in this talk, I want to look at the role of verb meaning and how this comes into play with valency changing morphology. So morphology that increases or decreases the number of arguments that are in a particular clause. I'm going to focus mostly on applicative morphology, which is a kind of suffix which adds a new object to the clause. These are talked about a lot in Bantu languages and actually in a lot of that classic work that Lutz mentioned earlier. And what I'm going to investigate here is how the meaning of the verb that the applicative attaches to determines the kinds of readings that you can get with the applicative. In the end, I'm going to talk a little bit, a very little bit I think, about the morpheme ick, which is a detransitivizing suffix in Kenyawanda. I'm going to look again at how the meaning of a verb can come into play with understanding the use of these forms. So just a little bit of background on argument structure and valence. Verbs vary in the number of arguments that they have and this often matches directly to the semantic description of these verbs. So if we see in 1a, we have the event of running, which involves one person, the person doing the running. And in this case, we have one argument, the subject John ran. In 1b, we have a transitive verb, two different arguments. John picked mangoes. This event involves two participants, John and the mangoes. And this sentence licenses both of those in the syntax as arguments. 1c is just another example with three arguments. One thing to keep in mind as we go through this talk, and that becomes a bit of a kind of nebulous thing, is when verbs describe an event that has multiple participants but doesn't tell you what those are. So for example, from English into a, we have John ate mangoes. So we're saying all of the participants that are there, John and the thing that he's eating. But in 2b, we can also just say John ate, and they're relieving you hanging on what that participant is. It's not there syntactically, but semantically, we know that it has to be there. He has to be eating something for that sentence to be true. Many languages around the world have morphemes dedicated to changing the number of arguments that a particular verb has. So one example, well-known examples are causative morphemes where you have an increase in the number of arguments and you add a new subject to the clause. So in 3a, we have an example from Japanese where we have an intransitive verb, the vegetable rotted, and then in 3b, we have a causative morpheme adding a new subject to the clause, making it two arguments instead of just one, excuse me, 3a. The suffix I'm going to focus most on today is called the applicative morpheme. Again, this is found throughout many of the world's languages. In bantu, it's a suffix that's added on to the verb. And generally, in many classic works, it's described as adding a new object into the structure of the sentence. So we have in 4a, this is an example from Kenyarwanda. The man wrote the story. This here is the verbal complex for, no idea what that is, for, it wasn't me. It was something else I stepped on just to be clear. We have the verb with two arguments. And then here we have this applicative suffix ir. Sometimes it'll come up as er, depending on some phonological factors that I won't worry about here. But crucially what this is doing is it's adding in a new participant and argument into this event. So we have in 4a, the man is writing the story, but in 4b, the man is writing the story for the child. And the child is this new applied object. And in this case, it's the beneficiary of the event. So often the applicative adds this new object and then adds some thematic information about that object. It's kind of the classic story of what applicatives do. One thing that I can make clear is that it seems to never be an agent or a theme. There may be reasons for that. I'm not going to engage with that here. But the roles that often show up with applicatives are beneficiaries, instruments, and locations. There's a very variety of others that you'll find, but these are the three that always seem to be there if a language uses the strategy. There are many, many, many, many syntactic analyses of applicatives in the literature, but almost all of them assume that the applicative has a very specific function. It adds this new object and adds some thematic role. And my kind of task today is to problematize that story and give examples of ways in which that's not the case, and then give an analysis of how we can more hopefully capture what applicatives are doing. Here are some examples from other Bantu languages. So we have, in Runyanbo, an example in five of a locket of applicatives. So we have the applicative ir, and then the locket of phrase there. In this case, adding a new object and assigning it at a location role. In six, we have an example from Ndebele with a benefactive applicative. So an event that's happening on behalf of the mother. And then in seven, we have an example with an instrumental applicative where the event is happening by the use of a stone. And again, in each of these cases, we have the applicative morpheme adding these new objects. So despite this general trend, as I kind of foreshadowed a second ago, it's been kind of pointed to in different works, different grammars, different descriptions of languages, that you don't actually always get an increase in the number of arguments. So here's an example from mostly locket of applicatives that some people have noted. And I'll come to examples of this in Kenyawan in a bit. But a lesser discusser point is that the contribution of the applicative is constrained by the meaning of the verb. So often these things are thought of as you take the meaning of a verb, you take this applicative form, you put them together, this is what you get. What I'm going to show here, based on work from Kenyawana, but you see this in other languages as well, is that the verb really says, okay, I have an applicative now, I have a set of options of things I can do, and what are those options is kind of the question that I want to tackle. And again, I'm going to look almost exclusively at Kenyawana here. In particular, I'm going to look at two case studies. So first is going to be different classes of motion predicates, so verbs that describe ways of motion. And I'm going to show that each of these different verbal subclasses of motion require a different thematic role for their applied object. This is evidence for us that the verb is what's in charge of defining what the thematic role is. I'm then going to talk about a class of di-transitive verbs where we don't actually see any increase in the number of arguments, but a change in the interpretation of one of the existing arguments of the non-applied verb. And from these two case studies together, we can propose that the contribution of an applicative is either a syntactic position, thematic material, or both. And then this typology can give us the entire range of meanings that you get in Kenyawana and ideally in other languages. I'm just going to keep running into this, I think. Okay, so there's a lot of detail that I'm about to give you, and we're going to go through a bunch of different examples, but the high level point here that we don't want to lose track of is that this case study is telling us that certain verbs require a certain thematic role for their applied object. So we're going to look at manner verbs, path verbs, and verbs of departure. I'm going to show that each of these has a specific thematic role. I'm going to go through these a bit quickly. So manner of motion verbs are verbs that describe motion and how the figure moves when doing that. So these are things like, in Kenyawana, quiruka run, gu tembera to go about or to wander, gu simbuka to jump. And in these cases, whenever you have the applicative morph, you get a locket of applied object, but it's specifically the goal of the motion event. So if we think of motion as a thing that happens, items start in a place, the source, they traverse some route, and then they end up at a goal, right? So in this case, our verb is telling us that our applied object must be the goal of the motion event. And crucially, it is not the case that it can be the source or the route. So this class of verbs, when you have a locket of applied object, can only have a goal interpretation. This contrasts with, say, path verbs. So these are verbs, the English equivalents of, excuse me, Kenyawana equivalents of inter, exit, descend. All of these have a route applied object. So crucially, the thematic role of this class of verbs is different from the manner of motion verbs, and this class of verbs can only have a route interpretation for the applied object. The cool thing is, I was really excited when I found it, is that you see each of the logically possible types of motion thematic roles. So you find a goal which we've seen, a route which we've seen, and then there's also a class of verbs for which you find a source applied object. So we have verbs like cross, to a light, to fly, and all of these have this interpretation where you get the source of motion as the applied object thematic role. And again, crucially, only that interpretation. So each of these verbs is saying this is my thematic role and my thematic role alone, right? Which is, again, very nice in that there's all the logically possible things are tested and we can see them. Here's a summary again. So we get a goal reading with manner of motion verbs like run. We get a route applied object meaning with path verbs such as enter and then we get a source with departure verbs such as cross. All of that is to say that this is telling us that verb class matters as far as the interpretation of the applied object because what verb you're using is conditioning what is possible as the thematic role of that applied object. So then turning away from motion verbs, we're going to set them aside. We'll come back to them in a bit. I want to talk a little bit about di-transitive verbs. So these are verbs. I'm using this term to mean verbs that describe three participants semantically. These are things like giving. Like I gave someone a book. There's me, the person who receives the book, and then the book. In this class in Kinyawanda, you often find that it can map two verbs with three arguments or two arguments and you find a variety of different subclasses, most of which are not relevant to us. So there's a class where the applicative adds a new beneficiary. Actually, I'll just go through the data as I talk about it. We find a class where the applicative adds a new beneficiary, actually giving you a sentence with four arguments. These aren't going to be relevant to us because actually this is the classic case of applicatization. You just add a new beneficiary argument. There's a class where there's only two arguments in the base form of the verb and then the applicative gives you a recipient. This is actually an interesting class in that recipient is only allowed as a thematic role in this particular class. So again, it supports this notion that verb class matters as far as what the possible thematic roles are for the applied object. We're not going to pursue this class much here. What I am interested in is this class, this Guterra class, where you have a lexically di-transitive verb with three arguments and the applicative comes along and doesn't change the argument structure of the verb. So in 16a, we have three arguments. We have Karimera, we have the rock, and we have Nhusi. And in 16b, we have that exact same number of arguments. Karimera, the rock, or in this case a ball, and Nhusi. The crucial difference between these two is the interpretation of that of Nhusi. So in 16a, the verb describes an event where someone is chucking the rock at Nhusi and he is probably not even aware that there's a rock coming at him. It may be done with some kind of malicious intent or a desire to harm Nhusi. Whereas in 16b, Nhusi is the intended recipient of the event. So this is like I'm playing catch with Nhusi and I want to give him possession of the rock. And these are really robust judgments with speakers. They really categorically put them in these two kind of categories for each of these classes of verb. And so it seems here that the difference between the two isn't related to argument structure, but it's actually the semantics. So one little kind of side trip we have to take is that it is logically possible that there is a transitive verb, gutera, with two arguments, and then that this kind of potentially problematic case is just built from that. So you're taking two arguments and then getting three arguments and actually there's nothing interesting, there is a change of valence and so my whole point doesn't stand. The next couple of slides I'm going to show that actually the three argument non-applied variant of this verb and the applied variant of this verb are related to one another. That's kind of a very... Is there like a mark right here? I think we'll be okay. So the reason none of this is actually a problem is that the transitive use of this verb means something slightly different from the di-transitive use of this verb. So in 17 we have the transitive use of the verb and it actually means something more like to play ball. It's ambiguous as to whether it's play ball with your feet or with your hands so it can mean something like kicked around a football or played something like American football. But crucially not throw. You're not displacing the ball and with the intention of putting it somewhere else. To get that reading you have to add some kind of dummy object so you have to add in this case ishoti, a shot, so who's he took a shot of some kind either throwing or kicking it. And just to really drive the point home, if you give a context so say I walk up to a crime scene and I see a knife and I pick up the knife but then realize holding this knife in this crime scene makes me look like I've committed the murder and I want to get rid of that knife as soon as possible. You cannot use the transitive use of this verb in that context. What this is telling us is in a context where you have to have directed motion you can't use the transitive version of Guterra. You'd have to use, if you wanted to use Guterra you'd have to use the die-transitive variant of this verb which is then evidence for me that that is the variant that relates to the applicative and so there's no change in valence. There's three arguments and three arguments. So all of this is to say if we kind of rope both of these case studies together that the verb determines the interpretation of the applied object in various cases and that there are cases where there's no increase in valency. I showed you here the subclass of die-transitive verbs but this has been seen in handfuls of other cases in other Bantu languages as well. So what I propose is that there are two components to the contribution of an applicative and crucially these two things can act independent of one another. So you can have a syntactic increase in the number of arguments namely adding an object or you can have the new semantic information about an argument or both. And the both case would be the classic applicatives and then these other combinations will be these other cases I've talked about here. I'm going to throw some lambda calculus at everybody but I'm going to try and keep it like lambda light. So to represent the semantic contribution of an applicative we want two things essentially. We want to say these are just functions, right? So we want a function which talks about some argument that is not part of the verb originally or we want a function that can talk about an argument of the verb, right? And that's the two different semantic contributions of applicatives. We have the difference, yeah, we have between is this argument part of the original verb meaning or is this argument part of the original verb meaning? And then we combine that with is this applicative increasing the valence or not? So in 21 we just have a simple lexical rule between the predicate argument structures where we have some number of arguments and then when we have the applicative we have an extra one or we can have a case where the applicative does not add a new syntactic argument position. So given these different possibilities we end up with three possible combinations of things. We have a case where you add a new argument position a new object and assign some thematic role to it great, that's what the classic view of applicatives always was we have cases where you have a new applied object so you have a new syntactic position but actually the thematic role of that is coming from the verb and not from the applicative and then in 25 we have a case where there's no additional syntactic position but something's being said about an argument of the verb and that is each of those kind of respectively. Ideally this is broader than just can you Rwanda this is a kind of possible types of what you should expect with applicatives the details in a particular language may vary of what works where I have some ideas about why you would find the kinds of variation you might find but the next stage of this would be to expand this study into other Bantu languages other applicative languages. Would this be a bad time to ask for the kind of difference between these three? Yes I could actually might help somewhere where I put that it's in my way. So in 23 we have the classic applicative case 24 would be Exactly I'll have some examples here in a second in 24 this is going to be the motion events where the verbs the one defining what the thematic role is and then in 25 it's going to be these die-transitive verbs where there's no change in valence at all. Again that's all for can you Rwanda specifically ideally this is a broader typology that you should have I'll show you here actually now. So we can walk through kind of a bit more slowly this first example so in this case we have with the verb Kuvuga which is to talk we have a new argument and a new thematic role so our classic case of applicativization so here we have the verb Kuvuga and then this is the semantic definition of what that verb means it's saying that there is one argument and it's linked to some event that argument is an agent and the event is a talking event which is all just a fancy way of saying that that's what the verb talking means right there's someone and they talk right and in this case the syntax here has just one argument. Then in 28a we have the applicativ that you would have in that case so this formula here I won't go through it all in much because the crucial point here is that this formula is just talking about an argument that is not originally part of the verb it's adding in a new argument outside of what was there originally and then the syntax is increasing the number of arguments so when we take 27a and we combine that with 28a we get if we go through the derivation in 29 which is boring we get 29c which is the combination of the applicativ and the verb meaning which is that there's a talking event there's some agent of that event namely the person talking event and then there's a description of the location of that event and then 29d we have two arguments again like we want and this would correspond to an applicativ version of this verb in 30a. The last one is talking in the house. Can you briefly explain what the applicativ is? This. What does applicativ do? It's applicativ. It's applicativ. Sorry. Sorry. Yeah. It's imperfect. Sorry. Sorry. Any other questions there? This might be a good time to make sure everyone's somewhat on the same page with me. So that would be the classic case of applicatives. What most people have noticed in the literature. In this case we have these motion predicates where the class of motion that you are is going to determine the thematic role of your specific applied object. And so what we have here is a slightly more complex story these Story, these verbs have more participants going on since, like I said earlier, motion always involves a source, a route, and a goal. So there's some details here that I will gloss over very, very quickly. But for this verb, cross, its applied object is going to be a source, which we'll see in a second. It'll combine with this denotation of the applicative, which the location is going to be an argument of the verb. So it's not talking about something new. And there is a syntactic increase. So essentially, what this is all doing is it's adding a new syntactic slot that the verb wanted to have but couldn't have otherwise is kind of a way of thinking of it. So essentially, the verb has all these participants. It has the full description of these motion events. But it only has two arguments to work with. So it can't tell you about all of these. So what the applicative is doing is giving it an extra position. The verb says, oh, thank god, I can now talk about this source, which is what we get here when we combine all the little pieces. And so we end up with a sentence in 35a, cutting it across the ocean from Mombasa. And if you were to take this denotation of applicatives and combine it with any of those other motion subclasses, you would get the desired result, but for the relevant, relevant, locket of applied object. This last case is where we have no increase in the number of arguments, but some additional semantic information added. So we have recall as verb gutera, which has three arguments. And we have this reading that Habimana through the rock at Caritesi. We have, let's see, the denotation there, which I won't go through. But in this case, we have a denotation that's talking about an argument that already exists in the meaning of the verb and no increase in the number of arguments. Nothing syntactically happening. And so again, when we combine all of these pieces, we get a denotation where the argument that's already present in the meaning of the verb gets an additional lexical entailment on top of it. Namely, in this case, the combination is that the goal of the not applied verb is a recipient in this case. And I capture this by combining the goal and the beneficiary. I have a bit of a story for how that works. But essentially, the sets of entailments of being a goal and the sets of entailments of being a beneficiary are both mapped to the same participant, and that participant then is interpreted as a recipient participant. And that's what we get in 40a, Habimana through the Rock, to Caritesi in this case. So then there's all these different kind of lexically determined subclasses of verbs and how they work with their applicative. But the question then becomes, what is an applicative? Broadly speaking, what is this form in a language? So I propose that an applicative is a marker of a paradigmatic output condition. And what this says is that not applied variants of the verb, so verbs that don't have an applicative and the applied variant of the verb, with the applicative, are related to one another. So you get the meaning of the one that's marked from the meaning that's not marked, and they're in a relationship with one another. To jump ahead a little bit, this parallels work on argument alternations, where an oblique phrase may be related to the direct object realization of a particular verb. And it's kind of via that paradigm that you know where the entailments are supposed to be. In those cases, the direct double object construction, say in English, will have a stricter set of entailments than the oblique frame. And so 41 would be this condition that all applicatives should have, which essentially says, when you have this pairing between a non-applied and an applied verb, the applied variant will have at least one internal argument. And the truth conditions associated with that internal argument are a strict superset of those associated with its corresponding internal argument in the non-applied variant. Namely, with the applicative, you should have more information about some internal argument. Either you say something more about an argument that's already there, or if you can't do that, you add a new argument and say something about that. That's going much quicker than I expected. So the conclusion from the applicative study is that verb meaning matters. This is important to determining how these valency changing morphemes work, which I think is very important, given that often these things are thought of as very productive operators that just spit out a new argument, new thematic role, end of story. This does is actually, if you look more broadly, isn't very surprising. We've seen from lexical semantic studies that verb meaning is very important to argument structure and does have all of these idiosyncrasies wrapped up with it. It just hadn't been brought into the realm of valency changing morphology. And so the next kind of step would be to look at other valency changing morphemes and see whether they have the same or similar types of verb restrictions. So I'm going to give you a little bit of an existential proof. I don't have much of an analysis to give you. But there's this morpheme ick in many Bantu languages, often called a stative, or a neuter, or a mediopassive, or a pseudo-passive. But essentially, it takes a verb and gets rid of the subject argument and promotes the internal argument in a way that somewhat resembles a passive, but not quite. So we have in 42 the verb comena to break, the child broke the cup, two arguments, and then the ick marked one here as ick is an intransitive. The cup might break. The cool thing about ick and what involves a lot more thinking on my part is that there are a variety of readings that you can get with this. Give in ick plus some other morphological and semantic facts. So you have this potential reading in 44. This cup might break. You can also get a stative reading, like if just someone can point to a cup and say this cup is broken, and you would use the break verb with this suffix. You can also get an encodive, this cup broke in the past, change of state, interpretation. And to kind of finish off this idea that verb meaning matters, you don't get it with every verb. You find quite a bit, actually, of lexical idiosyncrasy with verbs from different classes. So here we have in 45a, the bread is edible, perfectly fine, but you cannot get the reading that the bread, say, got eaten. So the encodive reading is impossible. You can't use it as the state of reading. The bread is all eaten. Even though semantically, those are all things that can be said, should be said, aren't in their own right problematic. There's something about this morpheme with this verb that's limiting the possibilities for what the readings are. So actually, what this could suggest as well is that sometimes the combination of valency changing morphology with verbs is just completely idiosyncratic. So we have a case where there's maybe productive application of these rules, which is the traditional way these have been thought of. There's all these verb class specific interactions that I was talking about here. So certain verb classes interact in specific ways with their chosen applicative, and every verb within that class will do that. But it's not entirely productive outside of that class. And then you may find that some of these morphemes are just completely unproductive. You find that actually, again, to come back to applicative, you do find it with a handful of verbs in kinyarwanda. They happen to all be unergative verbs, which I find very interesting. But the benefactive doesn't have a productive reading. So in 46a, you have guseka, to laugh. And when you use the applicative, it doesn't have the interpretation that, who is it? Karatezi is laughing on behalf of the child, or for the child's benefit, or for the child's pleasure, but that Karatezi is fond of the child. There's a couple other examples like this, where there's some maybe distant semantic relationship, but it's not entirely productive from the base meaning of the verb. So to conclude, I've addressed in this talk the question of the role of verb meaning in valency in changing morphology, using the Bonti language kinyarwanda as a case study for this. I've argued that the applicative can increase the valence of a verb, or add semantic information about an existing argument, or both, and that this captures a broad typology of possible types of applicative. And then I briefly, very briefly, pointed to some data from the detransitivizing morpheme ek in kinyarwanda, making a similar point that verb meaning may be a play with this morpheme. And that leads to kind of considerable questions for future research. So thank you very much. Great. It's time for comments or questions. Thanks. When you have the three-way typology of applicatives, would it make sense to collapse the first two by saying something like the increase as an argument and the detrol of the argument comes from a verb, and if there's nothing in the verb, then it's just unspecified. Potentially. The one reason I might not want to do that is, so I'll mention it briefly, that class of detransitives, where the recipient is the role. In those cases, I think for all of those verbs, I'd have to double-check, but for at least some of them, the applied object is optionally a recipient or a kind of classic benefactive. And so I think in that case, you'd want both kind of operations independently operating from one another, although I don't see your way of doing it to be unimpossible, I guess. But you might want to preserve that difference. I'm going to have to think about that. Yeah, I had a follow-up to that, maybe. Yeah, I was thinking that it sounded like you had an implicit argument there, right? Because you said that, if I remember, that sort of the conceit rules come from the verb. So it's somehow represented. For these detransitive cases. Yes, absolutely. That actually is the point. It is this class of verbs that does it because of those entailor recipients. When you say something, there is me, the all. I guess that's actually an empirical question of whether there's an entailment that there's a recipient. Yeah, so in disrespecting the fact that you're explaining it. But it seems sensible that that would be the class of verbs. Like eat, which even if you used them without an object, you would still think that there's something implicit in eating. Right, and with the verb like eat, you wouldn't expect that there'd be a recipient because. Yeah, exactly. And have something receiving that, yeah. So I think I agree. It's exactly because that verb strongly suggests that there's a recipient that you get there. And I have another thing for me. So when you mentioned 41, I think that was your constraint. I was thinking at the same time that it really looks like alternations. What people have said about alternations, say in English, would be double objects and prepositional data construction. So I mean, you could look at both dismantling the syntax of those in more detail. Because some people suggest that you have a null head there in the double object construction, whereas you have just over prepositional elements. And the null head might just be the thing that is spelled out in your type. The interesting difference, so yes, I'm essentially treating them as an alternation. The interesting thing is here, in some cases at least, but not in all. It's an alternation with nothing, which is a bit like in the non-applied variant you don't have anything to alternate with. So it's always added it in some way, which is somehow different. But basically, I think the same thing, yeah. No, it's basically an alternation, yeah. If you have one of these version verbs and you want to say the one that's not the inherent one, like if you want to say crossing to somewhere, how do they do that? They use what I tentatively would call a serial verb construction. So you have the verb to go. So if I wanted to cross to somewhere, you would say, I cross, I go, and then the place. But I don't know much about those. But I think there's a lot of cool stuff to be said, because I don't know whether you always use those or if you need how many different ways of saying it. I also don't know how you would add in a route, because to and from you have verbs for that, but you don't have a verb for being a route, I don't think. So there's a lot more to be said, but roughly that's how it's done, is with a serial verb-ish construction. To what extent can you say that this applies to back to languages in general? A little bit. So Lutz has some work showing that Swahili, for example, has cases where you don't get an increase in the number of arguments when you have an implicative, but instead some pragmatic kind of information added to that object. So it's a slight difference in that it's a pragmatic kind of effect, but I think that quite naturally fits in with the story that's being told here. A bit more needs to be said, definitely. And then the motion predicates. I briefly cited that literature. There are quite a few examples of languages. And again, there are usually just a few mentions in grammar or a paper where you get some kind of effect where a verb will have a goal, for example, with the locket in phrase, but it'll only be a handful of motion verbs, for example. So I think it's robust, but someone needs to do it, do a bigger study. So no one's taken this sort of three-category approach to the implicatives, the way that you have, with a kid or a mother that applies it to another language, or is there data for that in another language? There's not yet. If someone gives me the grant, I'm hoping they'll give me. I'm hoping. Yeah, I think that really, because there's a nice, in-depth, but also typological, kind of story that can be told here. So one of the big questions I have is, it really keeps me up at night, the degree to which this motion technology is replicated across languages, or if it's arbitrary, if it just happens to be the manner of motion verbs you can want to feel like being this way, but then in Bakusu, it's actually that they're all sources, or if in Bakusu it just happens to be that they're all general locations, or if there's regional reasons, or if there's, even at some degree, a human cognitive preference for these classes to link to these different roles. And at this point, that's a completely open question. Can I ask just one more on this thread? And have you seen anything similar in, like, applicativization in other languages, like a similar sort of structure? I haven't really. I have a colleague of mine at the University of Texas who's looking at Tepewan, which is spoken in Mexico, I forget what language family it's part of, which is, because I'm standing in front of a group full of people, but he has been doing some work there on that language, and I think there's some comparable things going on with the benefactors there. Other than that, I can't say that I know of anything, but nobody's also, I think, taking quite this look at applicatives in other languages either, and so I think it could be there, but not have been noticed or talked about in kind of a systematic way, like, language. Are you suggesting that they, like, are understanding the applicatives should be sort of redefined based on this particular finding? I mean, that sounds a bit bold of me to suggest that, but I think that the empirical evidence is that applicatives are much more variable than they've been talked about, at least in kind of the classic syntactic literature on, like, say, object symmetry, which is very much, you add a new applied object and then it has some thematic role and then here's a bunch of syntactic properties of that. And this is just showing that actually, the starting point is much further back. It's what is the applicative doing in this particular context and how does that interact with the meaning of the verb? So I have a paper on Bakusu that shows that the object symmetry fact, so this is where the, so when you add applicative to a transitive verb, you end up with the derived di-transitive, so you have three arguments and then a big question has been the degree to which those two objects are the same or different, so is the applied object a true object or is it more of an object? And what I found in Bakusu is that with a specific subclass of verbs, you get a different pattern from other verbs in the language. You're talking about the hierarchy of the objects, though. Essentially, yeah. And so there's been a lot of work looking at that in these language exercises, probably what most of the work on applicative has been in Bantu, but actually it seems, again, that semantics is coming into play to some degree and that it was ingested verb, say, in Bakusu, have a different pattern when, actually, it's what it calls it, but it's kind of the same, I think, ingested. Ingested, so eat and drink. Some, in some languages, like learn any kind of intake. I guess. So yeah, I think ultimately I would say that there's more here than people have been given credit for and that semantics is really a driving factor that hasn't been fully considered. Yeah, I was actually wondering about one of the verb types that you mentioned in the beginning because you said that, I think, you characterize cross as a verb of departure. Yeah. Could cross be a verb or was the... So okay, yeah. So the manner of motion verbs at the very beginning are a very robustly found class across other languages. The other two classes I really only kind of found here. So those classes are really just pattern name. Like they're only defined by being in this way. And I've wondered kind of a lot about why cross works the way it does, because again, any motion verb is gonna have a source radical. So there's no, it's not that verbs of departure care more about the source than verbs of jumping because in both cases, there's all three pieces. So... But I mean cross in particular seems to, I mean, you seem to sort of, it says something about what you're crossing, right? Which is the path or... Right, yeah. I mean, how you move on the path. So in the ocean that you're crossing before you select, right? So is it your question like, why is it considered a verb of departure if the primary argument that's taking is the verb that's crossing? Whereas actually independent of that, but that might be important. I mean I guess so it would differ from other, like what I was calling path verbs, like enter where I guess the thing is the goal. Yeah. The thing, the verbal object is the goal. So are you classifying them by what additional argument is out of that? Yes, yeah, I was kind of... Yeah, there's no independent motivation for... I think it's because of what they've done in public sense. Because motion's not really big. Yeah, so there is some work to, yeah, especially I think with cross. So in English there's been some work because there's this ambiguity within an into. So sometimes in has a reading of into. And I think punctual movements prefer the into reading or at least easier to get it. So if I say I walked in a room, you would probably think it was just a generic location unless there was a context in which I primed you to kind of know that there was a very quick punctual change. So that's something that very much could be a play here too actually, is the nature of the event. So yeah, can you go back to the local with cross? Yeah. And since this is the discussion, I think it has boots in it. It does, yeah. This is also in other languages being... Yeah, so... Oop, let me go. Oop, yeah. Boots, so do you have that as context? Because I think some people would have that as a separative suffix. So you could have, so fitting in with this discussion that I've just started as well, you could have the verb originally being something like am, or indeed amba. Boots being a separative extension which has now come grammaticalized to be cross from depart leave. So cross, but the cross essentially meaning having left in my own crossing is something. And that's now, presumably, no longer have the form. So this is now grammaticalized to include this point of... Tension. Tension, that's a good thing to check for. I don't know if it would... I guess there's also a hook in Guhagaruka. Yes, I mean the... The longer you're the match, the longer you cry. It's not clear for everyone's eyes, but I mean this seems to me to be, given the semantics and given what... And given it's all three, all of these verbs in this concept are given you have a book. And there could be a historical reason. So I would have that as morphologically complex. And how would that be glossed? Separate, so sometimes verbs like jump, so you're separating yourself from the ground. And different or... So the verb when you said jump actually, the verb jump also has the hook. I mean again, I don't know the degree, I don't know for sure any of this historically. But this would definitely be in a different class. I guess Quiruga also has the hook. So it would... This is, sorry, I switched to a different slide when you were talking. So all of the departure verbs have a hook, but actually now that I'm thinking about it, so do two of the manner of motion verbs. Which is to say that, which isn't to say that there's not something there, but I wonder what it is. It's a good point. So my original question is going to be, thinking beyond the day two percent of hitting, if you have an idea of how this all comes about. So of course I'm always thinking of language change. So if you have a story for why some of you are attendant of that thing. I don't. So, what's your first name? Paquirotti. Though she was here. Sara, yeah. Sara wrote her thesis on a historical story of Eve, which I have just been asked to read for a book review. So I haven't actually read it like cover to cover yet. So I think she has some ideas. I don't actually, I guess it's my answer. I think partly because I'm not sure, I'm not sure there's enough kind of detailed semantic work in this particular area, looking from this perspective on enough languages to know what the facts even are. Or any thoughts I have are extremely kind of just like, like something I was like, I don't know, thought of in the shower. It's not like something like, where I could really give an answer. I think. So I was thinking the other way around. So things that you presented perhaps as idiosyncrasies. Could not be. Yeah. That's some words in the past. But yeah. But I mean, that's a massive question. Yeah. I mean, I guess, so like when I brought up the completely lexicalized cases, like the being fond of the child thing, it is interesting that I have like three or four examples of those and they're all unergative verbs. So I mean, maybe there's something with, I don't, again, like, don't know what that would be, but there could be something with unergative verbs where that event just doesn't want to be productive. And so you get lexicalized. I don't know. I mean, it's a great question. But I don't really have anything interesting to say about it, unfortunately. How did you collect and go through the data after that? So it started like originally, years and years ago, I was a research assistant on a project doing machine translation. And I was annotating genocide testimonials, actually. And was noticing a lot of cases where you weren't getting objects where I thought you were supposed to. And kind of a variety of other grammatical things that just weren't coming up. And then eventually that project sent me to Rwanda to collect some data. And then I kind of just pulled the threads of things that I noticed in the speech text to get more kind of from there. So mostly interviews from this data, but the kind of noticing of something going on happened by looking at actual speech. Do you know if it's happens cross dialectically in Kenya Rwanda? No, because I don't know anything about dialects in Kenya Rwanda because there is a lot of them and people that are from even the same region will say things differently. So there are three different ways to pronounce the word for money and I don't know why there's three. Like the people I know who say these words differently aren't from very different places. So it's not, again, I have thoughts but nothing systematic and I don't do variationist work so I think somebody should go in and do a more detailed. And then that's also a country that's had a lot of internal movement because of its history. So I think you get a really interesting social-language story of people moving to different places for varieties of reasons or repatriating from different countries after many, many years in Uganda or Congo or something. So I don't know, but I think there's a lot of interesting and important facts kind of tied up with that. I was interested in how it ties in with your talk. Has it been, you would have seen some papers where people have characterized the use of a particular intervention. And the one that I can come out of my head are the obviously the benefacted one and the lobotin one, speaking about the dosy, you've got two others, the instrumental one, where you add an argument, which is a tool or something you're very interested in. And also sometimes you get the intensive one. Yeah. So the intensive I haven't been able to find in Kenya or Wanda. I had a chat with a friend who spoke Luganda and it seems like she had that in Luganda, but in Kenya Wanda I haven't come across that. In Kenya Wanda, the instrumental is a completely different morpheme and it's syncretistic with the causative. A few, that's kind of a regional feature of the Great Lakes area, so it's ish in Kenya Wanda. And I have a whole other paper on that actually. So it's, and actually, could have included it here because there's, so there's these two different possible readings of instrument or causative and the meaning of the verb actually can, in some ways, determine whether you get an instrument or a causative reading. So it's kind of the same story actually with the other applicative form in the language. I mean, it would be nice to think of that, but I just looked up my own data on the Chindamba here. And my example is one where it's the, called Diminence, where you need to cultivate. Mm-hmm. And the meaning depends what the semantics of the added object are, so if you add a person, you're doing it for them, and if you add that, how are you doing it really? Ah, yes, so that's worth saying as well that I've been a bit dishonest with you. But so has everybody else, I think, in this literature. So I kept talking about benefacted replicatives and lock-in replicatives as if they're separate forms and actually they're not. It's one morpheme that is the same. Which I think you're exactly right though, that the interpretation is gonna follow because lock-in replicatives are always gonna be marked with a little location on the applied object. I think one is clear, I think. Yeah, and then in Kinjarwanda at least, I can't speak to the other languages now, but in Kinjarwanda at least, it'd be disambiguated by whether there's a lock-in prefix on the applied object or not. And then with the instrumental slash causative replicative, there's, I think, some degree of the applied object, but also the meaning of the verb comes into play to varying degrees. Yeah, yes, so actually you're right that the applied object itself can determine an interpretation as well. Yeah. But you might expect that to happen less with your second task, right? Where the, so with those applied objects, which sort of just fill in something that the verb already has, you kind of, you would expect less semantic creation. If I'm still... So you're talking about the motion verbs here? No, I'm talking about those where the replicative doesn't seem to. So those that look like the English alternations, for example. Yeah, so I'm actually currently working on that class more specifically. And it kind of becomes, I basically can't wrap my head around it because you get these kinds of interactions, but then also sometimes a lock-in replicative. So it's like, it's really difficult, which then kind of overlaps with transfer motion verbs anyway, which is what often these doctors are about. So you can get like lots of overlapping patterns, I guess, to kind of summarize it a bit. So yeah, yeah, there is some interaction. It seems like you can get both types of applicative meaning, but they have slightly different meanings depending and it's still not clear what all of those are for all of the different classes. So which classes that would be interesting? This isn't those di-transitives. And so for example that... Oh, instead of throwing it out, can you throw it to them? Yeah, but then if you, you can also seem to get like a lock-in interpretation if it's not a person of like I threw it there that sometimes has it, so it... I mean that again is pretty much like in alternation with send, right? So it's actually that verb that's driving me a bit insane. And you can't send, I mean many people say you can't send the library to book, but you can send people to book. So there's a difference in the animal to know which... So in what sense of location do you have that perception? So that's maybe... Yeah, the tricky part is that you can, in some cases that we're finding, you can get like a lock-in-marked person in places where I don't expect them basically, but all that's... I guess at some point I could show you kind of where we're at with the di-transitives paper. Get your thoughts. I'm kind of banging my head against the table at the moment. I think again, your story is right. There's a multi-ethnic issue. So in Van Gogh's example, you can get, you know, jump or something at a place but you can also jump over people and you can't jump for someone and only jump over them. So even if it's a human, it won't be the location as well. But you can learn for someone. And, you know, it seems again, I think you know, you'll have to look at the detail of it. If you look at these sort of lock-in-marked people, okay, some of them allow interactive interpretation with the right object, but others are really resistant to that. And then, even though the context is very complicated but in fact, you still end up with verification. And it looks like again, it's basically semantics. Right. So even sort of a, even a very strong context couldn't sort of force jumping for someone else. I don't think so. I think then you have to go, except what Kaya said, to attempt construction on behalf of for the benefit of the proposition. That would work, but you don't accept it. That's nice. The other, well, I guess this is, it's kind of the opposite side of this, but so these lock-in-marked things kind of look like prepositions in Kenya, Rwanda. And I have evidence that they're not, but in some case, or I guess part of the evidence for that is that, so with, which verb was it? Kuvuga, to talk. With this verb, let me show you the applicative example, you have to have this applicative to give a lock-in-phrase at all. So you can't talk in the house without the applicative there. But this seems completely lexically, actually very unincorrect. So some verbs you can just slap the lock-in-phrase on there. No applicative and like everything's cool, but some verbs obligatorily have it. And that's where this, that transitive studies got confusing is that some verbs require this to have a lock-in and some verbs have a lock-in without the applicative. And it seems a bit arbitrary why that is, which from like a human learning language perspective is kind of blows my mind. Like I don't know how people would keep track of all of this lexical idiosyncrasy. But it seems that they must have made it. And so yeah, this verb, you have to have applicative to get a lock-in-phrase at all. If, like, can your Rwanda's use it so much? I mean, I just see there's quite a bit of variation. In speaker variation? Yeah. I mean, I've observed it. I don't have any, like, kind of. So does that present problems in like actually being able to determine these patterns of the language in general if some people might just be completely selecting these features differently? Yes. So that's why I made that little snarky comment where I said this is very robust of speakers because there is a lot of variation, but the patterns I've been hammering at, I've gotten really, almost surprisingly clear judgment, especially with like that Guterra case, the throwing to or at. To me, when I was like doing all of this, surprised that this was the contrast and kind of didn't believe that there was no change in veils, right? But speaker after speaker after speaker is very like, no, no, no, this is you threw it at him. He's not paying attention to yours being a jerk. And this one's you threw it to him like he wants to catch it. I mean, they really have strong intuitions about that. And that's been across every person I've asked about these verbs, which is like everybody I know. So did you go in with your semaphores already made? What's that? No, so it's kind of, I don't remember how I started looking at it. I think, so the motion ones, yeah, I don't know. As you go back and check with your previous interviewers mostly actually I should put, I don't think I ever made it this far. So there have been several people I've worked with over the years on Canada, but these people I have all been with like the people I go back to for this project especially. So while I have other kind of people consulting on things, I've run all of these and it's by at least this set of people and many of them, there's a lot of variety in age and in say expat status and things like that. Again, this is for people, so it's not like a systematic sociolinguistic creationist study or anything like that, but as to the degree possible I've tried to ask the same people the same questions the same ways to the degree again kind of possible. A really innocent and very ignorant question, would it have enough people? No, I don't think so. I mean, I think that like I was saying before, actually no, I wasn't saying this before, I said this like two days ago to somebody else. So technically I don't know before, but not like the way you would know about it. I think that this pairs really nicely with corpus type studies and I'd really like to do experimental studies on this. So I was briefly mentioning earlier how, I don't know and I don't think anybody knows why you would have these preferences for different parts of the path for example. And that could be genetic, it could be inherited from the earlier stages of the language, there could be kind of linguistic reasons for that or there could actually be cognitive reasons for that. The way to test that would be to put together some kind of novel, novel verb study. So these are used a lot in psycholinguistics where people make up verbs that sound like they could be part of the language but aren't. And so in this case they would describe different types of motion going different directions and then I would slap an applicative on there and be like, where's that going? Is the source of the route? And that would help weed out, rule out that as a possible explanation. Is there something bigger as to why or is this an idiosyncratic thing? So all of that's to say, I think for the patterns here it's enough in that it's robust across each of these people and then minimally this can describe their grammars. But ideally I think there's a bigger question of what these things are outside of Kenya one day even. And then one just funny question. The other sort of point about from a child language acquisition perspective. I think you could also think of that the other way around, I can't quite get my head around it but it's probably more important to know if I threw a stone at someone or to someone than if I spoke in a house versus whatever the alternation without the applicative is there, right? So like you're talking about the obligatoriness of the lock of applicative in some cases? Yeah, so hopefully the ambiguity that results from I spoke without the applicative is probably less. Those are only two examples that I can make. Yeah, I'm trying to think of more exact. So speak versus like throw out or throw to or something like that. Yeah. So in part that's the element of human objects. So that's something I mean, that's that's what we're going to talk about in a chapter it's being somewhat more salient because they tend to have human objects. And maybe that in your contrast is partly influenced by that. So I'm not sure that's the speaking versus the throwing of the entry. Yeah, so throwing a stone at someone is a perhaps more aggressive act than speaking in an arsenal. Yeah, it's also worth saying though that that kind of example was chosen simply because it really illustrates the semantic difference but that diatransitive structure can also be like I threw the stone at the wall, right? It doesn't have to be a human in that case. So it is just that it's a goal. Even in cases where it's a person though it's still a goal. But I threw the stone to the walls That I mean, I had asked about that and some were buried in a bunch of notes that I can't remember the pattern was we had gone through all these different diatransits and looked at whether you could give things to the wall in an evidence of things. So yes, there's more to be said. When I hear people talk about afflictives I sometimes think about this German prefix which I would like to think of as an afflictive and some of them just would throw I can now sort of as well say you can, what do you have in transitive? Transitive, you can add this prefix there not to all of us but to many of them. And so the throw for example the only accusative object that you can have I think would be the wall, I guess. You'd add the wall? Yeah. And so you, so I was thinking that maybe a sort of very exotic language from this perspective would be interesting to look at as well because those patterns So if you throw... So you can say I throw and usually throw takes an accusative object which is the thing that you throw but you can add this prefix there and then you say I, the throw and then the accusative object of that will be the wall for example. What happens to the object? What happens to the object? It has to be an oblique with Oh, interesting kind of instrument. The underlying kind of verbal object becomes the oblique instrument. Okay, that's interesting. So those patterns are fun and since they, they are very basic practical as well I think and you can have them with so throw would be one of them you have them with certain levels of motion as well. You can, so you can go or you can also go which means that you sort of then walk around an area and the area becomes the accusative object so kind of the goal in a way. Do you know if it happens in a way that you're just a gentleman? I think Dutch has a bit of this but in slightly different ways but I think it's something that people haven't really looked at. Yeah, the woman in my department when I was in Texas wrote a term paper on the Germans like it but I don't know how far she got her if she stuck with it after that. She was German so I think she was just opportunistically writing on something. I don't think it's easy to find an example. And it's, I mean the, it might just be an interesting sort of source of. Yeah, she definitely takes a look at it. Yeah, yeah. And again, it would kind of help to some degree with this kind of broader like what are the patterns outside? I mean, is it just that there's tons of like schooling and idiosyncrasy and kids just have to learn this or are there some kind of predictable things you get outside of this sort of, yeah. I think you may have mentioned this before but what exactly are you planning on doing with this? So yeah, I'm applying right now for grants to do this study on motion, motion verbs with lack of applicatives specifically. And so my goal there is to replicate this kind of semantically study in maybe two or three other languages which would involve a lot of sitting down with consultants and going through judgments. Having some kind of cross linguistic database of sorts. So looking through grammars and papers on a variety of Bantu languages from ideally throughout the continent. And then having again this novel word study in a handful of languages as well. So kind of pieces from each angle of it looking in depth at a small number of languages and then looking broadly at a lot of languages and then looking at these kind of what speakers invent when you give them new verbs that describe different kinds of motion. Still a lot too though. At this, yeah, at this point, yeah. Give a microsystem desire to just keep going back to that. No, it's not that. No, it's not that. And I mean, I think the micro variation in this case is really nice because the structures will look very, very similar that will be related to each other. And so if say, Kinyawanda and some other nearby Bantu language have completely different patterns, that's actually really interesting whereas it's not that Kinyawanda being different from Walpere or something wouldn't be interesting, but it's this, you can find so much variation just within Bantu that I think it's a nice starting point in that it's really, I mean, kind of comparing apples to apples and getting a sense of what the variation is within the subfamily. Is it enough to redefine what the applicative does just based on one language down there? So I mean, it's entirely positive. The nice thing I guess about this or one way you can think about this is I'm not necessarily saying every single language would have to have all of these different applicative types. You could probably and would probably find that you'll have some degree of this in each language and how this would work in each language would differ. I think verb meaning has to be relevant to every language. I mean, I really think if people looked at electrical semantics of applicatives in any language they would find some kind of stuff there. But minimally all languages, I mean the classic definition of applicatives that you add an object, add a thematic role. So all of them will share that, whether they have these other kinds of things I imagine they would, but it wouldn't be necessarily a problem. This position challenges that sort of basic premise of it, right? Because it's sort of talking less about adding arguments and more about... It's actually, I mean, yeah, you're right that in that my kind of conception of applicatives is a semantic definition. It's adding lexical entailment. So I guess from that view I would be redefining it, but it's entirely possible I guess too that there's a syntactic applicative out there in the world's languages where a language just uses it as a transit advisor adding a new object and in other languages you have this semantic effect of marking paradigms. And I think like Hannah said, there could be a historical reason for that, right? Like it's kind of conceivable that a language would have a productive operation that then would start to get lexicalized. Or vice versa, lexicalized kind of set of patterns that then just gets too unruly and so language starts to open in to have it be more productive. So I think all of these things can exist. The world will begin up for all of it, I think. And it wouldn't necessarily offend me. It all can fit within the same story. Because actually, technically adding a new argument and a new participant is increasing the lexical entailments of the clause. You're adding more semantic information. You're naming a new participant that's not there in the non-applied variant. So even in that case, this kind of classic syntactic case that's still actually captured by the semantic approach. So technically any applicatization is adding new information. Right, because if you throw it to the person you're throwing it at them for them. Yes. But even in this speaking case, so actually this is I think kind of crucial. So even in this case, where we have speaking in the hops, you have the applicative adding a new object position, assigning a thematic role to it, this is still increasing the set of entailments of the clause compared to just Kyle or Lasse spoke or talked. And so actually in that case, even though people have talked about this very syntactically, that is also satisfying the semantic constraint by increasing the entailments associated with the clause in a specific way. So does that mean that if you have the applicative here and then you have the house, which is like the argument adding the applicative that house is not an adjunct? No, it's not an adjunct, it's definitely an object. Whereas if you had it without the applicative, it would be an adjunct. Oh, if this verb you couldn't have it without the applicative. And I would even argue that in cases where you have verbs where you can get the locative phrase, it's still an argument. Yeah, because it can show agreement, it can be the subject of a passive. That might vary among languages, but definitely in King or Wanda. Are there any cases in King or Wanda where you take something that's an adjunct and turn it into an argument? I have yet to find it. So that's kind of another classic thing about applicatives is their thought often is this promotion operation where you have an oblique and then it's promoted. I haven't found any instance where you get a clear cut adjunct relationship with these. So I think they're worth looking at in their own right. Because your instrumentals are causatives. Yeah, so it could just be, I mean, other languages may have that kind of set of facts, it's not, but in King or Wanda specifically, it seems like you don't get that, that the applicative just does what it does. So for any kind of thing like in the house or last week or something like that? Yeah, so that's something I need to think through a bit more is how other kinds of information we come into play. Because it definitely seems like this needs to be a locket of some kind, this new information. So it's like they all add new information, but actually of a very restricted kind. Or like technically, if I negated this sentence, that would be adding an entailment to the clot. Like there's other ways you can satisfy this constraint that I don't want to deal with and wouldn't want to. So it definitely seems to be related to this argument and a particular thematic information about it. If there are more questions or comments then please let us think very much for this. Thank you very much. Thank you very much.