 It's a great pleasure to have Ken Sapphire here. Again, actually, you came two or three years ago. Three. Yeah, three. I hold the time of flies to give a very nice talk on his Afana project, which probably most of you know. It's a very nice web-based, database project, starting out with an Afran-African language, expanding to all kinds of other topics. So that's a nice thing to look at. And today, we're going to talk about Brab-Stansen-Egi-Mahawk and positionalists alternative. OK, is Serge-san now online? I'm getting him online right now. Oh, OK. So I want to say hello. I've never met him, so. OK, Serge. Yeah? OK, now another hello. Hi. I'm Ken Sapphire. Hi. So you'll be able to tell me when I'm misreporting Mamadou's judgments. But hopefully not too loud. Shut off his voice there. Wave. I do a little bit more adjustment now. I'll send you the hand-up by email from Mr. Lutz's email. So you'll have the hand-up. OK. So this is joint work with Mamadou Bassini. It's based on his judgments and the judgments of some of his close family members. This language is probably not significantly different, although it started an inter-family squabble by saying. So from Jola Banjal, which is a, I think it's, Mamadou's cousin wrote a book about that. But I found in looking through, I think it was his name, Anna-Cristian or something like that. Yeah, Anna-Cristian by the way. Yeah. Looking through his book, at least I didn't find anything that contradicted anything we said. So maybe I just didn't read it carefully enough. All right, so today I'm going to be talking about the language as Mamadou refers to it, which is Jola Egama, or I'll just call it Egama. And it's Atlantic language. For those of you who may not be familiar with it, it's one of the Bach languages. Now, these are languages that have a rich, agglutinative morphology. Those of you who are Bantuists are very familiar with this. And those of you who work on Atlantic languages as well. A number of them have this kind of rich suffixation. And the only question always arises is how do people construct meanings out of the various pieces that are in these rich stems? And if we think about what a fully harmonic language would be, as you see in one there, that would be a language in which the affix y, linearly closest to the stem x, forms a syntactic unit xy, a morphological unit xy. That's to say, some phonological process might apply to it. And xy is compositionally interpreted, such that those two pieces are added together in the meaning before anything else is added. And then you add something else on, and you get another further compositional meaning. And that would be the dream language, easily interpretable. You can completely undo the stem and know exactly how the parts added up. We are rarely so blessed. And we're certainly not blessed that way in Egama. Things seem to be out of order from the point of view of a compositional interpretation. And the things that seem to function as units aren't obviously things that one would identify as compositional units or as syntactic units. And the linear relations are such that certain elements of the stem that one might expect to be close to the root are far from it. So this is the kind of problem I'm addressing. I'm not going to solve all the problems of this very complex morphology. But there's a piece of it, a couple of pieces of it, that I think I have, Mamadou and I, have made some progress in understanding. And that's what I'm going to try and account for today. So at the heart of our proposal, it's really very syntactic in the way it's put together. The heart of the proposal is that there's a piece of the stem that moves and joins higher in the same stem. And so you're going to have a piece of the stem that is going to have roughly the order you might expect. And that piece is going to move to a position high in the stem, above positions where you would expect it to be, so that you'll have things at the bottom of the remaining stem that seem to be, they ought to be close to the root, but in fact, they're far from it. And that's going to be the reason why. And then I have to justify why one would make a move like that and claim I could get away with it, I've just moved it. Why not? Of course, you need evidence for that sort of thing, apparently. So we were forced to come up with some. And hopefully, you'll find it intriguing. OK, so that's the essential part of the analysis. We'll call that the inner stem movement analysis for reasons you'll see soon. Here's the, still catching my breath, four flights of stairs. The heart medication is just not up to it. OK. Tell me if you could say it, you know? Yeah, well, and for those of us who aren't, you're used to it, it's even worse, right? OK, so the order of march is on your hand out there. So first I've got to give you some basic facts. Then I'll set up the first puzzle, which I've sort of introduced, having material distant from the stem that should be compositionally closer to it. And then I'll make the inner stem movement proposal and defend it. And then I'll introduce two more puzzles, which are actually four puzzles altogether, which deal with the relative positions of argument markers and which is interpreted as unbenefactive and which is the direct object. And very briefly, there's going to be four of these things or three of them and some of them duplicated and so on. But these things are going to be roaming around the stem, but not that much. And the ordering of them determines certain sets of interpretations and precludes others. And the attempt there will be to explain why we get just the interpretations we do. The first attempt at doing that, I'm going to present a pragmatic proposal about how people, speakers might do it, that is very appealing, I think, and very simple. But it's also much less explanatory than a structural proposal, which I'll show, also derives important features of the actual orderings that are simply assumed if you take the pragmatic approach. And so ultimately, the structural approach can be more attractive. Probably we won't get to this. I could talk about some of the theoretical commitments of the analysis for those particularly interested in minimalist syntax or some reasonable analogs. And or I'll try to talk about how one might try to take this approach and apply it to some other languages, in particular to the Bantu languages, where we have some of the same kinds of reordering and some intriguing differences that, I don't know, on a stretch might be derivable from a slight editing of this approach. But that probably will be for the question period. OK, so from some basic facts, it's an SVO language. Subject agreement is really the only main prefix. There's one other little prefix that we don't quite understand. It's a little n thing that appears and disappears, depending on whether or not we have Irialis interpretation. If it's Irialis, the n seems to disappear. But we don't really understand that. And that's only with the third person, I think. Maybe it shows up with some of the other persons. I'm not sure. Third person singular. Anyway, we'll abstract away from both of the prefixes. Just assume that there's always subject agreement unless there's an infinitive. An infinitive is essentially a noun class marker. And I'll call it the infinitive. When it occurs as a complement to a verb as an infinitive, it's in complementary distribution with the subject marker. That's it for prefixes. And for the most part, I won't be talking about them further. When the subject marker is present, as it almost always is, the noun phrase subject does not have to appear, but it can. OK, there's no tense except for future morpheme, which is independent. And I won't have anything to say about that. But it is important to keep in mind that there are no tense morphemes. There are only a spectral morphemes. OK, there's a lot of suffixes. And so I'll sort of go through them briefly here. And then you'll see there's a summary afterwards. Well, closest to the stem, I'll work from the, closest to the root, rather, root for the root outwards. There are three affixes. There's the causative affix, which is probably not distinguishable from a kind of repetition affix that's directly attached to the root, and also the inherent reflexive attaches directly to the root. And since all of these attach directly to the root, only one of them could ever occur on a particular verb. One can't attach outside of any of the others. It's not clear that the causative and the repetitive are different, because they have the same phonology. So I'll just leave that open. I don't know why that particular portion of the, that particular affix is so close to the root. I don't have a story about that. But I have a feeling it has to do with the fact that there's a close argument relation. But the causative is productive, by the way. So it's not like it's a highly, lexically restricted affix. That's a common affix, productive. All right, I'll call that the innermost stem. So that's the root plus one of these affixes. The next portion of the stem will refer to it as the inner stem. And that is followed then by either a reciprocal marker or a reflexive marker, or both. The reciprocal marker has several other interpretations. I've called it the reciprocal marker, because it suits my purpose. But it has several other meanings. And if we have some time, we can maybe talk about those, because they lead to some interesting ambiguities. I only got a partial handle on those. But for now, we'll treat it uniquely for purposes of discussion as a marker of reciprocals, and like the reflexive marker, which also has one other kind of emphatic meaning, which I'll set aside also, these markers can occur in either order. And you could have two of each of them. I mean, if you had enough pragmatics, you could probably get more than three of these. And we're going to only talk about cases where there are two and where they represent arguments. That is to say, arguments of the verb, or benefactive arguments. And we'll come back to that. So they're the next two things. They can be there or not. But if they're there, then they mark the edge of what we're calling the inner stem. After that, you start to see a series of affixes that are strictly ordered. The first of these, we call former. It's kind of a used to meaning. Or on a noun, it would mean the former president or something like that. And then there's a kind of locative. And what Alain Christian, I think, called it centrifugal. So it's a question of whether something's coming toward you away from you kind of thing, where the action is coming towards you away. Followed by a habitual marker, followed by something we're calling a perfective marker, but we really don't understand exactly what it is. Because it seems to occur in some cases where what's going on doesn't seem perfective. Followed by negation, which has two alimorphs, depending on what I think the shape of the habitual marker is, I forget. And there's also an inclusive marker, which is earlier on. I think the ordering is in two. You see the full order of these things. The inclusive markers only goes with the first person to indicate who's in the conversation kind of thing. And finally, the passive. Now, there are certain co-occurrence restrictions. So that so-called perfective never co-occurs with passive object marker or a duplication. But the perfective, the passive marker would occur after negation, which is in a position where negation is in complementary distribution with the perfective. So that ordering is perfective before passive is sort of not real, because we couldn't test it. But after that, crucially, that marks what we're called the outer stem morphologically. And then after those things, you get the object markers if there are any. So those object markers then form the outermost stem. The outermost stem is the outermost stem, because the thing that attaches to that is a reduplicated stem. And that's the only place where the only thing that can occur outside of these markers, OK? So the way in which I've talked about these various units of the stem, as identified by the morphology, you see a kind of bracketing in three. That's not intended to be syntactic bracketing. That's just sort of how I've grouped the various morphological portions of the stem in number three there. And so you see, the innermost stem is the root plus the root attaching affix. The inner stem is that plus the reciprocal or the reflexive marker or iterations of those. And then the next portion is the inner stem plus the string of rigidly ordered affixes out to the end of those, which is perfective or passive. And then attaching to that are the object markers forming the outermost stem. And then the RED there, that's reduplication. If there is a reduplicated stem, that's where it occurs. OK. OK. All of these are then, or some of these, at least, are exemplified in the examples in four. So you could see, for example, in 4D, that the part that's, by the way, there's things that says pass there in 4A and 4C. That should be perfective. That's the wrong affix in the gloss there. That should be perfective into the pass. In C and D, you see reduplicated versions. And the stuff in the gloss that's in italics, that's the portion that's been reduplicated, the gloss corresponding to that. And as you can see, it contains the inner stem elements. All right. So one of the things we're going to have to talk about are the inner, sorry, the benefactive or indirect object. And the direct object and where they actually are an argument structure. And come back to that. But essentially, you get the summary now. So innermost stem, you see the exemplification in 7. The inner stem, you get the exemplification in 8 and A and B. The outer stem following that, then the outermost stem. So I'm skipping ahead now to what we get to the first stem puzzle, if you're looking your hand out. So you have generally the picture now, I think, of how rich the stem is, how there seem to be different regions of it, how they seem to fit together one inside the other, like a Chinese box. And we're left with this question of why is it that the stuff, things like passive, are very far from the root. And moreover, between the root and passive out here, you get all of these things that are sort of a spectral, like former and habitual. And so the puzzle comes in. If you think about it, Qin Kui has talked about the kind of clausal structure of adverbs. And one of the things that he points out, there was a clip for this. OK, it goes where? Can you hear me? Yeah, OK. All right, let's start with that one then. Yes, dramatic difference. Tell me if you can see that just well enough. OK, the essential idea is that there is a hierarchy of adverbial modification and a spectral modification, modal, and so forth. And somewhere down here, we get to the verb argument structure argument. So I always tell my students, this doesn't actually correspond to writing. It's a mnemonic for something I've just said. Moments later, it sort of disappears because it's illegible, OK? So that says verb argument structure. That's when I point to that, sort of like establishing a point for sign language. That's verb argument structure, OK? So that's presumably somewhere lower in the tree than things like various suspectual nodes or perhaps adverbial nodes. It depends on the adverb, OK? Some of them are very close to the verb argument structure and some are not. Repetition might be one of those that's very close. Whereas we talk about things like habitual, that's probably somewhere up in here, OK? Or used to, it's probably somewhere higher in the tree than you would think of things like I take a subject or I take a direct object. That stuff is going to be lower, OK? And also something you might expect to be lower, something maybe passive would be down here, right? And then you'd have something, some lower verb-like structures. Let's call them verb question mark, OK? And this would affect those arguments by, say, subtracting, say, the external argument or somehow putting it in Chomage or wherever you think of it. And yet what we seem to get is this portion acts like it's over here when this is fmr, OK? Usta, OK? This acts like it's here rather than there. So I'm going to say, suppose it starts here and moves here. If it starts here, then we get the kind of compositional interpretation that we want, right? And so now the only question is, all right, where do I get off saying that that thing moves here, OK? Here's how I get off saying that, OK? If you look at the diagram in 9, OK? You see essentially what I've done here with a little bit more and a little bit less, OK? There's, so you've got the subject up at the top in the agreement domain, OK? Which I'm assuming is outside of this whole, even the outermost stem, I think of the agreement as attaching, prefixing to all of that. And so what seems to be the case here is that stuff that's down at the bottom there, the inner stem, I'm claiming that that moves up to some place somewhere between that subject marker and former, OK? That's where it would attach. And then you'd see all the stuff former after that. I don't know why that happens, OK? I'm not going to claim I understand that. Suppose we say that it happens, OK? Then we're going to get the right ordering linearly and we'll have the underlying compositional structure that we want. So why can we say this? Well, notice that to do this, we'd be doing something like 11, OK? We leave behind some trace of having been there and move all that stuff attaching to what I've called the outer stem, OK? So what about that little t there? Well, in minimalist theorizing, all movement leaves a copy, OK? And then the phonology sorts out which copy is pronounced. And it's almost always the highest one, except in certain kinds of situations, which hopefully there's a good theory for. And then another thing that this modeling is used for is reduplication. What do you know? We have reduplication in exactly that spot, right? When reduplication occurs, that's where it goes. So the idea is that when this movement takes place, it leaves a copy below, which normally isn't pronounced. In situations where you want to avoid historical present or you want to say somebody really did this, OK? It's kind of an emphasis. That's when you use this reduplication, those two situations, so I'm told. And in those situations, the lower portion of the stem is also pronounced. So in a sense, reduplication is exactly modeled by the movement. It's exactly what you expect from this way of thinking about movement. And so that's my main justification for saying not only does this solve our problem with respect to this hierarchy and getting the right compositional interpretation, it also solves a morphological problem explaining exactly why the pronunciation of the reduplication is at this lower most point. The only thing I haven't, well, one of the things I haven't addressed is the position of the object markers. And I'm coming back to that. So that's the essential of the analysis, OK? And it's hard to see how another approach would unite these two kinds of things. The position of the stem, the compositional interpretation, so getting linear order, those two things together, and also deriving, just as a consequence, where the reduplication would have to occur. Because that is a consequence of this analysis. It's not an artifact of it, OK? OK. So that resolves the first puzzle we wanted to address. Why are the affixes out of order? At least that portion of the affixes. Why are they out of order? But now we have to approach these puzzles for interpretation. And let me explain how they line up. Well, we have multiple reflexive markers or multiple reciprocal markers or combinations of the two, OK? This is a language, I should say, first, where there is no applicative affix. So if you have, say, an intransitive verb and there's another argument after it, it's interpreted as benefactive if the pragmatics allows for it, OK? So you can get something like John died for Bill, OK? But it would come out as John died and it would be Bill. And the only way you'd make sense of it is to say, well, you must have done it for Bill. Bill's benefit, something, OK? So we don't see an applicative affix. A lot of the other languages in this region have applicative affixes. And so I'm going to take that as sort of an accident and assume that there's actually one there that just doesn't have any morphology. And that does a little bit of work for me, but I think you could probably do this another way. As long as we have the same argument structure consequence that I'm going to be assuming. All right. So that's how you could get a number of these things, even for verbs that are intransitive. And for transitive verbs, of course, you're going to get two. And if you've had, actually, there's cases where you can get three and even four, depending on how creative you are. But I haven't analyzed triples and quadruples for that effect, because I don't know all of the things working with Mamadou. For most of these stems that we managed to come up with, oh yeah, you could do that. But it would mean this. It's a remarkable range of ambiguity in many of these structures. Something actually isn't ambiguous. It stands out. In any case, let's take a look at a case where something isn't ambiguous. And that's in 13a and b. When the order of the reflexive followed by the reciprocal in 13a, that only has one interpretation. They praise each other for themselves. Now notice that for themselves, that's the benefactor. It comes first. And the reciprocal comes second. This has other interpretations, because the reciprocal marker has other interpretations that don't correspond to arguments. It can mean continual. It can mean simultaneously. And it can mean together. And I'm not addressing those meanings. And ultimately, we're going to have to address those meanings, but I'm not doing it today. Because it has to do with why it's in the inner stem, that's why I'll have to care about it. All right, so each other there in the 13a has to be the direct object, not the benefactive. And the reverse in 13b. So the order of these things matters. You can get either order, but once you have the order, the meaning is fixed. By contrast, when you look at 14, that's a combination where you see a reflexive or reciprocal in combination with an object marker. Now according to the ordering I've given you, that's always going to be the case that the reflexive or reciprocal will have to come before the object marker. That's a fixed order. And so because one of them attaches to the outermost stem and the other ones in the inner stem, you're never going to have the opposite order. But when you have that situation, it's ambiguous. You could get either reading. So they praised themselves for him or her, or they praised him or her for themselves, likewise. So in those cases, it's ambiguous. Now another situation that can arise is where you have two object markers. Now if the two object markers are of exactly the same ranking on the personanimacy hierarchy, such that you can't choose between them on that hierarchy, then their order is optional. But the first one, when they're optional, is always the indirect object. And the second one is always the direct object. If, however, the personanimacy hierarchy intervenes, and that is listed in 15, then it's always ambiguous, because that's a fixed order. That means that an object marker that's human must always precede one that's non-human. First person must always precede. First or second person must always precede third person animates before inanamates. And when all those things are set aside, then say third person plurals will precede third person singulars. So you have all of those ordering things, and those are fixed. But if you manage to get two object markers that are a tie on the scale, then their order is optional, but their interpretation is fixed. When their order is fixed, their interpretation is optional. So that looks like a very simple way to sort of pull these apart. And so I'll just skip ahead to the generalization in 18. So fixed ordering relations between argument representing affixes A and B typically allow for ambiguous interpretations of underlying semantic role composition. But if surface ordering of A and B is optional, then surface AB requires A higher than B, and surface BA requires B higher than A. Now I'm assuming that indirect object and benefactives are higher in the argument structure than direct object. And this goes along with a whole line of thinking that has been done on applicatives, and especially in Geronimo Grammer, but not uniquely there. So now here's my straw man. But it's a pretty tough straw man. I mean, it's just like, you might look at this and say, yeah, that's the story, and that's in 19. The optional order has rigid interpretation because the speaker's choice can signal the intended interpretation. But when the order is rigid, the intended interpretation cannot be distinguished by morpheme order. So the listener must continue to entertain two possible interpretations that the speaker may have intended, and usually in a pragmatic situation, you're going to know which one was intended. So if you know what orders are fixed and what are not fixed, then presumably you can do this kind of reasoning and you will get the right result every time, as far as I can tell. All right, so that looks pretty appealing. I mean, why not stop there? Well, the reason I don't stop there is because it doesn't explain why the orders are fixed or why the particular orders are the ones that they are. And one would like to have an explanation that did that and got this result. And that's what I'm going to try and do next. And so essentially, in the structural account, the way you get ambiguity is that you have two underlying structures, but only one surface order. I mean, that goes right back to the very beginning of January Grammar in the 1950s. And I'm applying that kind of a strategy to explain which cases are ambiguous and which ones are not. The ones that are not ambiguous are where there's only one underlying structure that could correspond to the overt morpheme order. And I'll show that the pragmatic strategy, appealing as it is, is less explanatory than the structural approach when all is said and done. In part because the structural approach already relies on the inner stem movement, which we've already seen has done some of the work for us that we might otherwise not have anticipated, but not only for that reason. So I am assuming something about the argument structure of applicatives that's very important, or at least the relationship between benefactives and direct objects. And you see that in 20, where I'm assuming that there's some kind of applicative head. It's higher than some kind of verb root head. And that to form a normal verb, what happens is, especially if there's a benefactive argument, is the verb root raises to the applicative node, joins to it, and then that together raised to the little v node, which identifies the root as verbal. And that's your typical verb root. That means, however, from the structure you see in 20, that the benefactive argument is higher in structure than the direct object, and this is going to be crucial. That's the part that I need out of the applicative being present there. So even if you assume that there's no applicative marker, unless you can recapitulate this much of the structure in terms of what's the commands, what, that's what I need, crucially, to do what I'm going through here. However, the way I'm doing it, big v applicative adjoining together to little v is crucial for certain assumptions in minimalist movement theories. And if you're interested in those details, then it has to be done that way, and it has an interesting consequence if it's done that way. All right, so now I'm going to make some other assumptions. And some of these might seem unobjectionable or even elegant, and others might seem to you, if you're not familiar with minimalist theorizing, they might seem to you exotic. So hopefully, if they get too exotic, you'll stop me and say, that's ridiculous, and I'll say, you're wrong, and then we'll continue. OK, so in 21, so I'm going to assume that the object markers, the reciprocal marker, reflexive marker, they're all argument suffixes. Again, again with the proviso that the reciprocal marker has some other meanings, which I'm ignoring. And that means that since they're suffixes, it means that in the phonology, they're going to be right adjacent to some verb step. That just has to do with the morphological requirement they have. But by saying they're argument suffixes, that means they originate in an argument position. I've given you the argument structure. So if a particular object marker or reflexive marker is going to represent a benefactive argument, then it has to originate in 20 in where it says Ben. And it will be in complementary distribution with a full DP in that position. So if you get a full DP, then you don't get these markers and vice versa. So a full DP would be like John or the man. If you get that there, then you can't use an object marker as far as I can tell. All right. Now here's where I start getting more particular in my assumptions. I'm assuming that the object markers attach to a phrase. The phrase they attach to is something this edge. This phrasal thing that just above the argument structure, essentially. The way that I'm cashing that out in this particular theory is that the phrase projected by the little VP. The little V is the thing that tells you you have a transitive structure, especially if there is the subject argument. If you want to call it that, it says EA here. That means external argument. That is introduced in the specifier position of little V for all transitive verbs within this theory. And so the little VP contains the argument structure of the whole unit there. And my claim is that the object markers, this is where they land, right here. And they attach to phrased. That makes them different from the reflexive marker and the reciprocal marker, which are otherwise the same except they attach actually to the V0, little V. They attach to a morphological unit. That's the difference between them and the object markers. And hopefully, that'll get us all the consequences we need for their distribution. So in D there, you see what we have, how a complex verb would be formed, as I've described it. First, the big V would adjoin to the applicative. And then the applicative, the V applicative would adjoin to the little V to make a complex V. And it's to those things that the indirect object and direct object reflexives and reciprocals would be adjoining to. And so they have a different landing site in the structure than the object markers. That's going to be crucial because the object markers are going to escape this thing. That's how they're going to be stranded at the end. Because what's going to happen when we move the inner stem, the inner stem is going to be that complex V with the reciprocal and reflexive markers on it. So when that V moves, it's going to strand everything else. Well, what's left? Everything's moved out. Well, all that's left is going to be the object markers, if there are any there. OK, so turn the page. I don't know if you're turning the page. It's 23. I'm out of sequence for you because I have a little more stuff on my page. So in 23, you see what happens if the direct object and the indirect object happen to be object markers. Now, if they're not object markers, and you're simply going to see in those positions first the indirect object and then the direct object, that is the unmarked order for a benefactive and direct object is a benefactive first, the element second. So what will happen then is that these two things will adjoin to VP. Now here's the crucial part where I appeal to some minimalist technology, OK? Both of them are heading the same place. How do we predict what order they end up in when they both move to the left? Remember, the order is crucial, right? Because if they're equally ranked, the higher one better be the indirect object. So how does this work? Well, there's a theory put forward by Norvin Richers of MIT some time ago. And it's one of a set of proposals that are called shape preservation. When you move more than one thing, there are structures where they move, so to speak, in tandem. They always seem to end up in the same order after they've both moved. They don't change orders. And these affixes I'm arguing are like that. And so I'm taking one of the theories that makes this prediction, and I'm applying it here in a way that is typical of it. And it turns out to get the right result and one more besides. OK, so the way this theory works, shortest move is the economy principle on movement. And what it says is when two things are going to the same place, you make the shortest move you can make first, and then you make the next move is the next shortest move you can make. Well, the shortest move to the adjunction to the VP would be from the indirect object, because that's higher in the tree. So that's going to go in a joint first. But notice I've got the direct object marker underneath it. How does that happen? Well, the next move, of course, is to move the direct object. But if it only has to get to the edge of the VP, it doesn't have to go outside of the higher one. It only has to go as far as the other one. In the adjunction theory that's typical of this theorizing, those positions are equidistant from each other. And so, but they're not equidistant from the launch point. From the launch point, the closest place you can move to and get out of a little VP is to tuck in under the indirect object. This has been employed, is mostly used to account for superiority effects in Slavic languages, for example, where you have multiple WH phrases moving to the front. And the ordering of these phrases is fixed in certain ways. The higher one that's moved has to end up left most. And then the next one has to be the one after that. And you can't get the other order. So those are languages in which you have multiple WH movement. They're all going to the same place, just like this. The ordering is preserved. The underlying sequence is preserved. That's exactly what's happening here. So I'm applying it to exactly the same kind of case with the same kind of reasoning as in the superiority literature. OK, so that means that the natural order will be when you extract two direct objects that first will be the IO and then the DO. Now remember, looking at 23, when this hits the phenology, the thing that that indirect object will be next to will be a verb. Because the verb stem is going to go all the way down to here. And then when the phenology comes along, these guys are going to be sitting here. And they will attach to the verb stem because they are suffixes. That's what we said about them as morphology. And moreover, the top one will go first simply because it's closer, and then the other one. Although I've set them up so that they both left were to join here, the actual order there doesn't matter. All that matters is one's higher in structure than the other. And the first one would count as the first suffix attached. The second one will be the second suffix attached. OK, so we now have the fixed order of 2 OMs, object markers, when they are of the same rank. OK? Now, when this happens, since we've preserved the argument structure order, that means the interpretation will be fixed. There's only one interpretation here. The higher one has to be the indirect object. So that gets us why the higher one will be the indirect object, and the lower one will be the direct object. Now, if the animacy hierarchy intervenes, it's going to mess this up. It's going to reorder these affixes. There's two ways to approach that. One is to simply say, well, there's a morphological fact and you reorder the affixes according to the hierarchy. And that will work for my theory. There's another theory that says that that also is a structural effect. And Mark Baker and I have written about that actually looking at Bantu double object constructions in a workflow paper of two years ago. And I think that theory will work here, but requires more work than I'm prepared to do today. And I'm not sure it's right. So let's just stick with what we can say harmlessly. And nobody has a better theory of those things. And so whatever the hierarchy is that leads to fixed orders, that's going to disguise the structure. If it disguises the structure, that means we have two potential underlying structures, therefore ambiguity. So we get the ambiguity in the case of two object markers. Now let's go to the case of the reflexive and reciprocal markers. When the reflexive and reciprocal markers move to attach to the complex V, remember, they're attaching to a morphological unit. They're not attaching to a phrase. So you see in 25 what will happen there. Looks like I didn't get it all on one page. Well, maybe you have it all on one page. You do. OK. I was more careful with what you were reading than what I'm reading. So you see the complex verb there. It has the reflexive marker and the reciprocal marker. That's where they have to attach by hypothesis. Now the only question is, how do we ensure the order IODO? Because remember, this is fixed. You can get either one can move first, or sorry, either one can be closest to the verb, but the one closest to the verb is always the indirect object. Now if you look structurally here, structurally, this is reversed from the case we saw earlier. You're looking at that complex V. The highest thing in it is the direct object. And the lower thing under it is the indirect object. So why does this come out differently? I just had this nice story about shortest movement. But notice in the shortest move diagram, if you're comparing 23 and 25 now, those adjunctions of the object marker work to phrases. And therefore, one of the phrases is C commands everything. The one that tucks under it tucks under its C command domain. Now in the case of adjunction to the V, the first one adjoins to the V. And by the shortest move, that's going to be the indirect object. Now notice, though, for the direct object, it's not shorter to go and tuck in under the indirect object here. That's not a shorter move, right? Because it's not the same kind of hierarchical structure you have in 23. The shorter move is to attach above, OK? And in fact, there's another principle which I won't introduce here that I've argued for in a 2010 paper in biolinguistics that would, in fact, rule out that kind of tucking in here, OK? It will rule in the other one and rule this one. It actually exactly did this long before I ever thought about these cases. So what will have to turn out to be the structure is the DO is higher than the IO. Well, now, again, you hit the phenology. The IO is closer to the verb. It'll attach first. The DO is further from the verb. It'll attach second. The IO had to have come from the indirect object position because it was attached first by shortest move. And so it's predicted that the order should be IO, DO, and unambiguous, OK? And so the optional orders where either one of these elements could go reciprocal reflexive, their actual structural representation is always the same, OK? And so there's only one structure. There's only one interpretation, OK? There's only one more thing left to explain. Because now we have the two optional orders lead to fixed interpretations because you can read them off the structure, OK? The case where the object markers are obscured by the animacy hierarchy gives you two underlying potential structures which you can't tell from the morphology, so this ambiguity. And now we have the case where you have one reciprocal marker or reflexive marker and one object marker. That was another instance where we had fixed order, OK, because one is in the outer stem and one is in the inner stem. And the result was ambiguity, OK? So why should it come out ambiguous? Well, you see in, I think I have 25 more than once, don't I? OK, well, in the second 25 and 26, OK? You see these two structures. And I'm going to stop shortly after this because we're running low on time. Essentially, what's happened here is that first you've done the affixation of the RCM, reciprocal marker, to the complex verb, OK? And then you've done the extraction of the object marker to join to the little VP, OK? Subsequently, you will have inner stem movement, which will take that complex V and move it higher in structure, OK, stranding the direct object at the end. And so you'll have the order, verb stem, reciprocal marker, other stuff, object marker, OK? And if you get re-duplication, you'll see the inner stem again after that. Those are ambiguous. Why should they be ambiguous? Well, if you look at 25 and 26, second 25 and 26, what you get is the same surface order for both cases, although the structural orders, the structural representations are different, OK? So in the first one, we took the indirect object and joined it to the verb. Remember, since the direct object and the indirect object are going to different places, shortest move does not apply to them. That only distinguishes, you do the shorter move if they're both going to the same place. These are going to different places, so you don't have that. So the indirect object, if it attaches to the complex V, is not slow. The direct object will attach high. Remember, there's no morphological difference between the object markers when they represent benefactive or direct object arguments. The direct object will come out last. The reciprocal marker will come out close to the verb root. In the opposite case, you have an object marker taken from the indirect object position and joined to the big, the little VP. And the reciprocal marker is joined to the verb. When this plays out in the phenology, the reciprocal marker, you get, again, the same order. You'll get verb, reciprocal marker, object marker, OK? But the underlying structure will be different, and that's why you get ambiguity every time. So I could go a little further here. There's an interesting kind of climbing structure in eagema, where if you have infinitive complement, all of this stuff plays out on the higher verb, OK? You can either leave the stuff downstairs or you can move it all up to the higher verb. If you move it up to the higher verb, all the same interpretive properties hold. All you have to assume is that instead of targeting the lower verb, the RCM and the RFM simply target the higher one, and that the OMs target the higher little VP. And then when that thing reduplicates, it strands the lower infinitive, as you would expect, because it's only pronouncing repeating the verb of the higher verb, not repeating the lower verb. Or you could have a duplication of the lower verb, in which case it has the same form that normally would if it was not embedded, OK? OK, so all that works just as it should. So the conclusion for eagema is this. We could have been happy with the pragmatic story and said how speakers on the fly interpret what these things could mean. And that may even be how they do it on the fly. I don't know. But there is a structural theory that not only explains why there are the ambiguities that there are and why the fixed orders result in those ambiguities and why the optional orders result in fixed interpretations. We can explain that in structural terms plus explain why it is that object markers are stranded at the end. Why it is that these reciprocal and reflexive markers travel with the inner stem and solve the compositional problem for the interpretation, OK? And why there are the sets of ambiguities that there are, it's all structural, OK? And the pragmatic account simply makes no commitment to how the orders come about, all right? Doesn't explain that. This theory does. All right, so I think I'll stop there and I'll take questions and then maybe we can talk about a few other things if there's time. Thank you very much indeed. I'm sure there's many points of issues being raised by that. And floor's open for questions or comments. I've explained everything. There's just no way to raise an objection. Yes, please. Thank you for this talk. And I'm still digesting, so my friends are playing on the topic. You raised yourself the question of how people do it on the fly. And I think for me, that's the most important question because in, we know from research on Bach, languages and so on and so forth, give me a language is that there is a huge amount of argument edicts, which in this case means object ellipses. So no marker. Yes. So all your data look, and you mentioned them, pretty much elicited. So these examples, you're very unlikely to get on the fly in any of the languages in the area that I'm familiar with. So what does that mean for your account? Well, I'm not sure exactly what it is you're claiming. You're claiming that these are not things people say? No, these are things that people can say in particular context or in the absence of a particular context. Most likely in the absence of a particular discourse context. No, sometimes we had to create a context in order to see if something was possible. And in other cases, we created a context that simply wasn't possible. But I'm just wondering how your account would deal with object ellipses, which increases the ambiguity considerably. Well, I don't. There's something that's struggling. So far as I know. But is that object empty ellipses or ellipses of these little markers? That's what I'm trying to. Both. OK. All both. If both are missing, that simply isn't, except for a few verbs, isn't what I know for megama. So you might also note that this is a language that doesn't have the applicative morpheme. And so there are other hints that you might get if you have an applicative morpheme and you're missing those arguments that you might be able to recover, which you can't in this language. So fully alighted noun phrases for transitive verbs don't seem to be possible. I mean, with no object mark or no direct object. And maybe Serge can tell me that, oh, there's lots of those. But I tested for those. And we don't seem to allow it. Well, we have been in personal communication with Serge, Sanya, and with Alain Christon Versailles, who told us exactly that. Oh, OK. Well, that's not what I am. All right, so if there is ellipsis, then I have to figure out what it would mean for the particular theory. I mean, I would presume that you would have the same structural underlying relations. And so if an object marker was suppressed, say, then in those situations where the surface order ought to have given us a strict non-ambiguous interpretation, then an ambiguous interpretation should result. That's what I would predict. And if that's frequent, then a lot of those cases, it's going to be ambiguous. And you're going to need the pragmatists to figure it out. But structurally, it doesn't challenge the account so far as I can see. Because what really matters is in those cases where you see these things, there are certain things you can't do. And we have a structural account for that. So simply saying, well, the pragmatists will help you figure it out, it doesn't do the job there. It's the things that you're not allowed to do, because the syntax doesn't let you. That is what I'm accounting for. I think your strawman, it's not a good strawman. Because I don't think it's a pragmatic story. Pragmatics is maybe here and there. But what is interesting, I think, in your pragmatic account is that really it's a morphological account. What is crucial about it is that you are saying, well, there is no syntax, there is no structure. I mean, that's the way. In the strawman account, yeah. But somewhere you have to state which orders are fixed and which are free. And that statement seems, that sounds like morphological statement, not a pragmatic statement. Pragmatics, then, you have to assume that morphology, but what it does, the pragmatics, that's all very likely. Oh yeah, I'm assuming that. I mean, assuming that the pragmatic story, I mean, objects of morphology are not objects of pragmatic theories, as far as I understand it. Pragmatic theories about strategies and discourse reference, and so on. And so you wouldn't refer to a piece of morphology as an aspect of a discourse theory. It's rather something in a morphology that the pragmatics can exploit or not. And so I'm assuming that any pragmatic theory simply assumes that there's some ordering story that goes on in the morphology. But wherever there's more than one possible outcome, the pragmatic story says, oh, then it's ambiguous. That's all I was claiming about that theory. And it seems to me that a proper pragmatic theory shouldn't go any more deeply into the actual morphology because they are apples and oranges. So the way that I'm presenting this, I mean, again, as I said, one could always appeal to the pragmatic theory and say, well, that's how people actually do it in a given situation when they're not sure. They don't compute all this other stuff, and they just do that. Well, maybe that's true, but we don't know that they don't compute all this stuff. And I wouldn't undersell what the human mind can do in these environments because we don't know. But to me, it looks like it's a good environment for lexicalization. So I think, in a sense, this is a situation where you maybe can have your cake and eat it. I said, it looks like there's structure going on. I mean, there's details in which I don't have a mechanism in principle. I think you're absolutely right. You want a structural account, but still there may well be a case for saying that in actual speech context. What people do is they just lexicalize these chunks and go for best guesses. Well, that's why we chose a particularly bizarre cases sometimes, so that people then wouldn't be accessing things that they normally have to put it together. And the question is, could they put it together? And when they couldn't put it together, I mean, because Momadou says he had lots of long conversations on the phone with his uncles and saying, well, but suppose uncle that you could do this was a situation where he says, oh, yeah, then you could get that. But then it would have to mean this. And we went through several years now of figuring out what his uncles are thinking. So the point is that what I described as the pragmatic account needs a place to start. This theory tells you that the place to start is also tells you a lot about where you finish and that you probably don't need the pragmatics to do quite so much work, that a lot of it's being done by the structural relations. Now, in terms of lexicalization, I do think there are aspects of lexicalization that I have to appeal to. For example, I don't understand why the causative is in the inner stem. And that is something that seems to be very much driven by some kind of deep and evil morphology that says causatives have to be. It makes sense that it would be down in this sector with the rest of the argument structure since causative does that. So things that add, there we go again. Fishing, there we go. Got one. Things that add structure would presumably be down there. So cause might be down there. But I don't feel that confident with that. Especially, see, a lot of the languages where you see that kind of thing or something you wouldn't expect to be quite so low, it's usually like in the Bantu languages, you get these affixes that are not frequently used or they're not productive and they might occur closer to the stem. I think Jeff Good has talked about some of these kinds of cases and Larry Hyman. But cause is not one of those. I mean, not this cause. I mean, if you're talking about the short causative of Bantu, now that's something else. But that would notice, here's the Bantu story if it ever gets off the ground. And I don't know if it'll ever get off the ground. But here's what could happen. Passive in Bantu, this is famous carp thing that Larry talked about, Larry Hyman. Namely that there's a tendency in Bantu for the affixes of the verb, the extensions to line up causative, applicative, reciprocal, passive. And that there are occasional blips in the direction, usually, of getting a more compositional interpretation. And he put these as a kind of opposition, whether try to put it in an optimality theory terms where you might rate the template higher in some languages and the composition higher in others. He called it mirror. And those competitions would produce variation. But that required assuming that there was something deep about this that doesn't seem conceptually necessary in any sense. I mean, why should this be attractive in some conceptual sense, such that Bantu languages would want to have it? You say, OK, well, there's a historical tendency. But that's, you know, but a lot of languages don't seem to get past it for certain kinds of constructions. Why should it be there? But if you notice this piece, OK? That's CA, causative, applicative. That's also part of my inner stem, OK? And passive is out here on the end, as it is here, OK? It's just possible that some version of an inner stem movement could be moving these things, actually, sorry, the wrong one, right, from there, from somewhere lower in the stem to this position. Now, in the best case, there would be some re-duplication kind of evidence. And I haven't found any, OK? But maybe people who know a lot more about this can think of. There's causative, not reduplicative. Yeah. The causative reduplicates and the size to be affected. Well. Is it reduplicative now, or is it, like, a longer form of the causative? Yeah, see, I don't think that's reduplication. We're all out of the forms, yes. Yeah, see, that one is also out at the end, OK? So that's another part of the puzzle. And in fact, I believe, is it after the passive, the short causative? Is anybody here? It's before it. It's before it. Yeah, I'm not really sure, OK? I know what kind of verbs to test, and I haven't got a chance to do it, because you take verbs like some of the group. What's it supposed to move? You take verbs like grooming verbs. And some of the two languages, they don't take any object, and they don't take any morphology, and they're interpreted as inherently reflexive, OK? And in many of those languages, it's only a few of the grooming verbs that do this. Some of them don't have any of them that do it. And so you'll have something that will mean, you know, wash with no object, OK? But I'm going to just look reflexively. And. Like English. Yes. But some of them will then take the short causative, OK? And when they have that, then you could get John washes himself. Or you could get the, you know, you get the ethics. Or you could get, you know, a direct object of any sort, OK? So what this does, it says, you know, go ahead. Have any kind of direct object you want. OK? It's transitiveizing. It's not causativeizing because it's not like John then undergoes washing when he did it before. All right? So it's not, it's something that really makes it transitive. And the passive of these is the one so I want to look at. John was washed. In the languages where I can set this up. And Little Pusa was one of those, and I just worked with somebody from that language. And so I just have to get in touch with them to see how it comes out. So whether this, sometimes there's a certain invocation and so on and so forth. Sometimes it just enters the nearby consonant in the direction of politics or something. But so that would be an interesting case. In any case, what it would mean for this. If that stuff is here. Right? The weird thing about the short causative is that it's way at the end. I mean, you take this and you actually need the courtesy. OK? Because those languages that have the short causative it's way at the end. And so some of those languages will have both of these. Some will only have one or the other. OK? And some of them will have both of them for certain indirect interpretations of causation and others for direct causation. So there's a lot to be unraveled there. But the idea would be that maybe the piece that moves is there. Right? So the stuff that you normally expect to be close to the verb is not close to the verb. OK? And it's one other little glitch that might be really cool. That's this. OK? The difference, one of the differences between Bantu and these languages, Islamic languages, is that you don't get these object-marked reclining effects. None of the Bantu languages I know about allow you to take those things out of the lower cause and the object-marked reclining effects. Some of these are acceptable cases. Not my own language. Now, suppose that the difference is that in Bantu when you're having it forming your inner stem at the OM and it's in there. OK? It attaches to the complex verb, not to the VP. When this stuff moves, it's going to line these things up. OM, RFM, verb. You'll get exactly the order that you get. And there's just some evidence that this is always closer to the verb than the direct object. So we have two objects and one of them is a reflexive marker. It's not a good second one. That's a reflexive, right? And so if you can get two argument markers there and one of them is reflexive, it's going to be this over. Now, of course, the reciprocal markers shows up in different places. There's more stories to be told about that. Actually, the long story we have about that is motivated by completely different considerations, but that's not for today. So this could be a way of thinking differently about the Bantu verb steps. And trying to recover some of the compositional structure that right now is very hard to detect in the way that some of these analyses have gone. And we might be in a better position than simply relying on, well, some of them do it that way and some of them do it the other way. There might actually be some rationality to why this particular ordering might emerge. And it might emerge because there's an example. That's the hope. You know, can you go back to Eugiman? The reciprocal marker has other functions. Yes. So it can mean simultaneous. It can mean together. And it can mean continuously. Now, the one that I find most disturbing is continuously because that nearly doesn't fit into the story as I have told it. That one seems to be kind of a spectral. I mean, you know, it's sort of maybe fallen with repetitively. So it could be one of those very low adverbs. But it's not at all obvious to me why that's so. The reason I say that it's all one affix is because the ordering restrictions on it are exactly the same as, you know, snot doesn't have a different sound. So there's no reason to think that there are, there's more than one affix there. There's just more than one meaning for what looks like one affix. It's the same story with the causative and repetitive that I gave, right? Because we couldn't really tell the difference. So I don't know for sure that there's a real difference between them. In the other case, the causative and repetitive, I think there's some restrictions on when we have causativized transities. So the tendency to interpret those as repetitive rather than causative. But I think we also eventually discovered you can get causative meanings out of those too. But it required a little bit more context. So I don't know why continuously would be one of the readings of this thing. And if you look at the many meanings that reciprocals have across Niger Congo, that one usually doesn't go with reciprocal. So, I mean, certainly simultaneous and together those often occur where the range of meanings for reciprocal markers is expanded, but not continuously. I had no idea why that. So I don't have a theory of it. But it's important for you that it's an optimum position? Well, I don't know. What I have to now test are triplets. So when you have, say, orororo, right? So when you like, gusaloororo, right? That would be something like, he's making a face. So they are praising each other for their benefit continuously, right? That would mean, can you get that? All right, so... Well, yes, take something like gusaloororo. Yeah, it's hard because it's 13. Which one? 13? What was the number? Oh, I think he's frozen. No, no. We'll get him back. You're talking about 13. 13? Well, there's two 13s. Which one? It's not Sergio's fault, it's Edgar Ohm's fault. So he couldn't... Well, it's an odd situation, right? Yeah. Yeah. So that one, I had to give a context to make sure that we could get it. So they... Hi, so you're back, right? So you're asking, I think, about 13... Was it 13a or 13b? 13b? Oh, okay. And 13a you did not get? Yeah, well, I only know that, you know, I have to go with my guys, right? So it's Mamadou and his uncles. But you speak a completely different language. I think Sergio has some comments he'd like to make. What's the difference between A and B? Yeah, some comments, that'd be great. Yes, 16... 16c. I thought it had to be first. So the all... Yeah, the all is the us, right? And so it's... The all, O-L. Yes. Yeah, well, O-L, yeah. Right. Right. So there were cases where... What's the third-person plural is you? I forget. What's third-person plural for people? Yeah, but I... You're saying that? Yeah. Well, I'm looking at 16d. And the plural comes first. 16d. Sorry guys, we have to leave last week. Okay, I'm afraid we're running out of time. But hopefully we'll... Somebody will do this in person. You can send me e-mail. Yeah. Please send me e-mail, okay? If people are feeling... Yes, I think it's under... But before that, thank you very much for... Thank you.