 Our speaker today is our very, very, the camera guy who joined our department in January to work on the project by HSC, which is about possessors for two years. But his talk today is not about possessors at all. The specific topic is about languages which have differential object marking and in which the exponent of differential object marking is homophonous with date of case. And so we find this in a number of languages, for example, in Spanish, which is on the first page of your handout in examples one and two, where in 1A we have a transitive clause with a definite direct object. You'll be able to see the book just subject web object, nothing special going on there. But in 1B, when the direct object is animate, we get this differential object marking element. So we have, I see the woman and we have this appearing here. In 1C, we have a die transitive construction in which the direct object does not get differential marking, but there is a recipient argument and we see that this recipient argument is introduced by an element which is homophonous with what is glossed as differential object marking in 1B. So we have, I give the book to the woman. So that's the basic phenomenon. So we have this element, which can appear with what I usually call direct objects and indirect objects or recipients. Two shows an example, another die transitive with the verb return now, which shows that if we have an inanimate recipient or goal argument, like library, as in 2, we still get this, even though we have an inanimate argument now. So there are some differences in, so basically direct objects have differential marking, that's what the name says, but indirect objects are introduced by this ah in Spanish in general. Spanish is not at all the only language which does something like this, though most of the data I'll be talking about will be from Spanish, European Spanish. But I'll, other languages that have something similar are Hindi, Kashmiri, Bask, at least some varieties of Bask as well. Most of, most Romance languages with which have differential object marking show something, show this kind of homophony, but it's also found in some other languages outside of the Indo-European family. And the way that I want to approach this question is, or I want to ask the question where this, where this homophony comes from. And what I'll try and do is sort of have, think about two extreme hypotheses. One is that, that homophony comes is because of syntax, meaning that when we have this ah in Spanish on both direct objects and indirect objects, this is because, so this, the hypothesis could be that this is because they express the same grammatical relation. So whatever object is in that grammatical relation is spelled out with ah. On this view, we would, we would explain the homophony just by saying that, yeah, these are syntactically the same kind of syntactic object and that's why they get the same spell out. This is what some people have suggested, in particular for Spanish, but the logic or like, ah, Romance in Spanish, but the logic sort of is supposed to hold for other languages as well. Ah, if this is true, then we would expect that, if, so if we're looking at one grammatical relation, we would expect ah, objects with differential object marking and indirect objects to behave alike in syntax. Ah, so that's one hypothesis. The other hypothesis is that the homophony is just simply due to morphology. These languages happen to have ah, this differential object case marker which just accidentally is syncretic with dative case basically. So nothing in syntax tells us that these two things should be identical, but they come out as identical in ah, in morphology. So the idea here is that we're dealing with two distinct grammatical relations like we're used to from languages which don't have this homophony, say a direct object and an indirect object, but these are spelled out in the same way. On this hypothesis, ah, we don't necessarily expect these two kinds of objects to form a natural class in syntax because we're looking at two distinct grammatical relations. And what I'm doing in this talk is to apply a couple of tests to structures involving both and see whether direct objects with or without differential marking and indirect objects patterned together or differently. So that's the basic idea. And we'll look at, we'll look a tiny bit at some different object positions in the clause. We'll look at how these arguments relate to passivization, reduced relative formation, controlling secondary predicates, nominalizations and some other aspects. So the basic claims that I'm trying to, or that I want to make are that in Spanish, Hindi, Kashmiri, Basque and probably some other languages, zero-coded direct objects and objects with differential object marking form a natural class. They are both direct objects and should be analyzed as that and they are distinct from indirect objects. And towards the end, I'm not sure what has any handouts left. And towards the end of this talk, contrast the Spanish, Hindi, etc. data with data from English and Gandhi which allow a kind of argument structure alternation in certain verbs which I think provides some more evidence that we're looking at sort of, there's different ways of grouping grammatical relations basically. Okay. If we turn the page, we can start with syntactic tests. So the first thing in 2.1 that I want to talk about is just very briefly about object positions. Luis Lopez in his book on differential object marking in Spanish and some other languages discusses some evidence for a structurally lower and a structurally higher object position. So in example three, he provides a certain context. It's about enemies and taking hostages and so on. And so he says that there's a context. What did the enemies do? The enemies delivered X to Y and Z to W, but los enemigos no entregaron a suijo a ningún prisionero. So the enemies did not deliver any prisoner to his son. Now he points out that in this sentence, you can either have the R, so you can have differential object marking on any prisoner, or you can not have differential object marking on that object. Both sentences are fine, but only when you have differential object marking can the direct object bind the possessive pronoun in the recipient. So that would give us the meaning that the enemies did not deliver any prisoner to that prisoner's son. So he suggests that that reading is only possible when we have differential object marking, which suggests that the object with differential object marking is structurally more prominent because it can bind the possessor. Even though this is not reflected in the word order in example three. Hindi is another language which shows different object positions in connection with differential object marking. We see this in 4a and b. We have a di-transitive with the verb send. So we can have the word order RAM to Anita a letter send, in which the direct object letter doesn't have overt case marking. But we can also have 4b where we have RAM, the letter, core, Anita to send. So we have, again, we have the homophonous core on both the direct object and the indirect object in this language, and we get this different object position. So one could say, okay, so maybe the spell out of this homophonous element comes about because of a structurally high position in the clause. So whenever you have a direct object that is in a higher position than its base position basically, maybe that's why you spell out the same, an element in the same way as the indirect object. And this seems to be what's happening in Spanish and Hindi, but these are not by no means the only languages which do this kind of thing. So loads of other languages have distinct positions for objects independently of whether they have differential object marking or not, and independently of whether this differential object marking is homophonous with datives. So in 5a, b, c, d, I just have a couple of examples from Canada which has differential marking as well as well as distinct object positions but no dative differential object marking overlap. So in 5a we have something like i, adverb well, then book and read. So we have the object book following the adverb, and this means something like I enjoyed book reading. This is a non-specific reading of book. This means something like there were multiple reading events and it doesn't have to be the same book that's involved in there. However when we keep that word order, so when we have an object following the adverb and then add an accusative suffix, we get a specific reading of book. So we get something like I enjoyed reading a book and this is about a single book. Now we can further sort of modify this example by having the direct object precede that adverb. So we're looking at a different position there and now we only get the specific reading. So with or without differential object marking. So this language allows, has different object positions as well. The higher position is sort of associated with a more prominent meaning in that it's specific. But this doesn't mean that they're expressed in the same way as dative. So I haven't shown you an example with a dative there but the language has a distinct dative suffix. So basically those higher object positions are not something that we only find in this dative differential object marking languages. That's the point here. And languages like German which don't even have differential object marking also show different object positions and so on. So that's the point about object positions. The next point we can look at to see what kind of grammatical relations we're dealing with or how these grammatical relations behave is passivization. So in 6 on page 4 we have the passive counterparts to the active sentences in 1. So 6a is a Libre Fuebista. The book was seen. In 6b we have La Mujer Fuebista. Both of these are theme passives. So we passivize a direct object, a theme or a patient. It becomes a subject. And note that this happens independently of whether the direct object would have triggered differential object marking or not. It happens with the inanimate and it also happens with the animate object. But 6c shows that Spanish doesn't allow passivizing a recipient. So we can't say La Mujer Fuebista, La Mujer Fue dada el libro with the intended meaning the woman was given the book. So we have this test here which shows us that we can passivize direct objects in Spanish but we can't passivize indirect objects. Turning to Hindi, we can look for the same things. In 7a and b we have first an active sentence in 7a. Ram will carry Anil. And 7b shows that the direct object can be passivized. We can get Anil will be carried by Ram. Straight forward. In 8a we have an active, die-transitive sentence. Ram sent Anil the necklace. And 8b and c show that this structure can be passivized in both ways with the dative in the first position or the absolute in the first position. Mohanand suggested in 8b the dative is actually a subject but what's crucial is that the recipient retains its dative case here whereas as we've seen in 7b themes don't retain their accusative. So again even though the core suffix is homophonous for both types of objects they behave distinctly in passivization. Moving on to page 5 there is possibly a slight complication here in Hindi because some speakers of Hindi at least allow retaining an accusative under passivization. So 9 shows that according to Mohanand we can say something like Anil accusative was carried by Ram. So a passive direct object does not necessarily become nominative it seems. Now Rajesh Bhatt told me that at least for him there is a difference in meaning between the accusative preserving and the non-accusative preserving types here. So in 10a and b we have the same lexical material Ram in earthquake was killed. So Ram was killed in earthquake but Rajesh Bhatt told me that when the accusative is present on the subject for him there is a more active reading here. So while 10a means Ram was killed in an earthquake 10b means something like Ram was murdered in an earthquake. So there is a difference in meaning here with respect to how agentive the thing is. The crucial point is that for indirect objects that doesn't seem to be this possibility. So direct objects can lose their accusative and often do lose their accusative and for some speakers that's the only possibility whereas indirect objects always retain their dative. And this is so we basically see identical things in Spanish and Hindi here that direct objects whether they trigger differential object marking or not they can become nominative subjects whereas indirect objects in these languages cannot do that. The next test we can look at refers to reduced relative clauses which actually also involves a kind of passivization and the structures I'm thinking of here are shown in 11a and b with English examples so something like 11a the book given to the woman where we have a reduced relative clause headed by a theme the book or 11b where we have a reduced relative clause headed by a recipient so the woman given the book. In English both of these are possible and perfectly natural but not both of these so we don't find both of these structures in Spanish and Hindi. So in 12a and b we see sort of by now familiar examples we have a Libro Bistain Lacaya the book seen in the street that's fine we can also say La Muchera Bistain Lacaya the woman seen in the street also fine so again independently of whether we have whether we would have differential object marking or not these noun phrases can head reduced relatives. 13a shows that we can't do this with a recipient we can't say La Muchera dada a Libro the woman given the book that's ungrammatical and 13b just shows that this is not something about that transitive because we can have a reduced relative headed by the theme so we can say a Libro dada a La Muchera so it's not about that transitive it's about relativizing the recipient. Hindi now shows the same picture as Spanish so now the recipient just sort of or the indirect object patterns just like in Spanish in that in 14a and b we can say something like the woman seen in the market so we can have a reduced relative headed by a theme in 14b we can have the book given to the woman so we can also have a die transitive reduced relative headed by a theme that's fine but in 14c just like in Spanish we can't have something like book given to the woman so the woman given the book is out in Hindi as well so looking at reduced relatives we again see that direct objects seem to form a natural class whether animate or inanimate specific or nonspecific to the exclusion of indirect objects the next thing we can look at is controlling secondary predicates again I'm giving some English examples to illustrate what this should be about so in 15a we have a sentence I have seen you drunk and as is sort of often said in such transitive sentences the subject or the object can control the depictive secondary predicates so in I have seen you drunk drunk can refer to both the subject I or the object you context will help you disambiguate what's going on here but looking at die transitive constructions this is something that is much more difficult for the indirect object I have given the book to the woman drunk and here drunk wants to refer to the subject the book will not be drunk anyway and in 15c with the double object construction in English I have given the woman the book drunk I think it's still sort of more much more salient to have drunk refer to the subject so we can look at this in other languages as well so on the top of page 7 in 16 we have examples from Spanish we can start with something like with an example with an inanimate direct object my mother bought the washing machine broken what's very practical here is that the depictive secondary predicates in Spanish shows what its controller is because it agrees with it in gender so in 16a it could also be that Rota refers to the subject but that's sort of an irrelevant meaning here but in 16b when we have Juan as the subject and Maria as an object we can tell by the agreement on the depictive predicate what its controller is so 16b and 16c are a nice little pair because we see a very similar surface string but we see that they don't really allow the same pattern so in 16b we have the predicate which takes a dative argument let's talk to so in 16b we get Juan Lea Blua Maria Boraccio Juan talk to Maria drunk Juan being drunk but we can't have the depictive predicate agreeing with the object so we can't say Juan Lea Blua Maria Boraccio that's out so basically the dative argument cannot control the depictive secondary predicate in 16c if we change the predicate from talk to to find both options become available so we get Juan Lea Lea Controa Maria Boraccio referring to Juan but we can also say Boraccia with the predicate referring to Maria so a direct object can control the depictive secondary predicate an indirect object or a dative object cannot 16d and e show examples with domestic violence and di-transitives I'm going to skip those show the same picture basically now I'm not going to talk mention Hindi here because I don't have the right kind of data but a variety of BASC so some dialects of BASC have differential object marking where the differential object marker is again homophonous with dative case so we have an example in 17a including a from standard BASC which has a dative on grandmother the second word in the clause so we have something like I have carried the child to the grandmother happy and the indexation on happy shows that in the sentence I have carried the child to the grandmother happy it's possible that the subject is happy that the child is happy but it can't mean that the grandmother is happy so again we see that the recipient or goal argument here cannot control the depictive secondary predicate in 17b however where we have I have seen you happy and you here is marked with the homophonous differential object dative marker the secondary predicate can actually control can be controlled by the direct object so in I have seen you happy it's possible that you are happy so once again looking at depictive secondary predicates across languages these two languages at least shows that direct objects pattern alike irrespective of whether they have differential object marking or not they can control secondary predicates while indirect objects cannot for the next couple of examples I'll only give examples from a single language but they sort of add to the picture in general so the first point on page 8 refers to nominalizations in Spanish 18a shows a sort of straightforward transitive sentence with differential object marking so we get something like the dog captured Juan we have an animate object we have a proper name so that triggers differential object marking that's fine when we nominalize this verb so we make out of captured we make capture we get something like 18b la captura de Juan por el perro fue sorprendente so the dog's capture of Juan was surprising so what happens here is that we as the object of a nominal sort of of the nominalized version of capture we don't have differential object marking anymore but 18b shows that la captura de Juan is ungrammatical so basically what we see is that the differential object marker changes to the genitive when we're in the nominal domain we can extend this to diatransitives as well so in 19a we have a verbal diatransitive Maria le entregó el paquete a Susana Maria delivered the package to Susana and what we see here is that we have an inanimate direct object differential object marking on the package we have a dative marker on Susana because it's marked as the recipient argument when we nominalize this structure we get la entrega del paquete a Susana so we get again the direct object being expressed by a genitive in the nominalized construction but the recipient now retains its a, it retains its dative so once again we see that whether an argument triggers differential object marking or not it's the object of a verb so this distinction is neutralized when we look at the nominalization and in the nominalization it becomes a genitive irrespective of the animacy, specificity of the object however when we're looking at recipients or goals which are marked with a this a is retained in the nominalization so it looks like we again see direct objects patterning together indirect objects patterning a different way the next example is from a southern Italian variety called palizese which also has differential object marking so animate and definite objects have to be marked with a now we don't see this a so much on its own in these examples because it confuses with the definite determiner u giving rise to a so in 20a we have something like I saw the book no differential marking, just a definite determiner in 20b we get I saw or child which indicates that differential object marking has happened here and if we turn the page to page 9 in 21 we see that we can have, we can get this or as well in a di-transitive construction so we have something like I gave the money to the child and to the child is or child just like in the direct object version however what's very cool about this variety is that its dative has a genitive alimorph which should be greek influence if I remember correctly and this is also called the greek dative basically that you can express a dative case using a genitive alimorph so this is shown in 22 which is just like 21 but rather than expressing the recipient with dative case we also use genitive here now the cool thing is that this genitive alimorph only targets recipients but it doesn't target direct objects so 22b where we have the genitive alimorph for the direct object is ungrammatical, that's just not possible so again we see that now we have sort of alimorphy which targets indirect objects or recipients but not direct objects or themes Spanish, doubling in some varieties of Spanish we see clitics which are sensitive to the case of the argument so we can have distinct dative and accusative clitics, other varieties don't have that distinction so this is not a very strong argument but yeah not bad to mention this the final point that is interesting to make I think is comes from Kashmiri now which also has differential object marking which is homophonous with dative case and one thing that is particularly interesting in this language is the case on pronominals on pronoun objects because direct objects which are personal pronouns are only marked with dative case when that person is more prominent than the person of the subject so if you think of a prominent scale where a first person is higher than second person which is higher than third person then when you have a first person subject and a second person object you get on the pronoun object but if you have a third person subject and a second person object then the second person object is more prominent and you get dative examples show this a bit more clearly so in 24a and b we have I am teaching you versus he is teaching you so the only thing that changes in these examples is the person of the subject we have a first person in 24a and a third person in 24b and correlating with this change in person we see a change in the case of the object so in I am teaching you the object U is expressed in its nominative form whereas in he is teaching you the object is expressed in its dative or differential object marking form now crucially this thing only applies to direct objects when we have indirect objects they are always dative and the person of neither the person of the subject nor the person of the object influences this dative in any case so in 25a we have he will hand you over to me where he is third person so it's like the least prominent possible person with respect to this case marking alternation and the direct object U is marked dative or in its differential object marking form and so is the indirect object me both are dative in 25b when the subject is the most prominent I am handing him over to you more prominent than both the direct object and the indirect object the direct object is nominative morphologically unmarked whereas the indirect object is dative so indirect objects keep that dative irrespectively of the person of the other arguments so 26 just shows that this alternation is only found in the imperfective aspect so in the perfective we don't get this we don't get this alternation on direct objects but indirect objects are still dative that was a whole load of data basically what we have seen here is that looking at passives in Spanish and Hindi, reduced relatives in Spanish and Hindi, secondary predicates in Spanish and Basque as well as normalizations and some other patterns these things sort of converge on one point showing that direct objects whether they trigger differential marking or not behave as a natural class with respect to these constructions whereas indirect objects always pattern in a different way so I think that these data are fairly clear for these languages so I now want to briefly talk about English and Khanti to illustrate how in some languages it is possible to express themes and recipients as the same grammatical relation so this is a kind of counterpoint to what we've seen so far because both English and Khanti allow in a fairly straightforward way to show that themes and recipients can be the same grammatical relations so that they pass all of these tests in the same way in the same way but sort of differently than Spanish Hindi etc so the main idea here is that it's well known that English has a di-transitive alternation for some verbs including give so we can have in English give can have or can appear with what is often called the prepositional dative construction I gave the book to Mary where the recipient is expressed with two or we can have the double object construction I gave Mary the book a lot has been written about this there's some references here on page 11 and one way of thinking about this is illustrated in 27 and 28 and you don't have to take these structures too seriously there's sort of more to illustrate a crucial property of both constructions so in 27 we have what I call the recipient prominent structure so in some sense the recipient is more prominent the recipient gets accusative in some languages the recipient will agree with the verb and so on we'll see examples of all of this in a theme prominent structure it's the theme that is accusative the theme that can agree with the verb and also arguably in a higher position than the recipient maybe so in languages which allow both of these we get some more we get some more freedom than in Spanish and Hindi with respect to some of the tests we've seen so in 29 we have two di-transitive examples from English I just mentioned that we can have I gave the books to Mary or I gave Mary the books straight forward now both of these allow passivization but in both of these we can only passivize the more prominent argument in the sense of 27-28 so I gave the books to Mary allows the structure the books were given to Mary and we can kind of tell that we passivize the prepositional-dative construction because the recipient still has its prepositional-dative whereas the books has become the subject here so that's 30a in 30b in Mary was given the books we can sort of see that we passivize the double object construction because there's no this can originate from I gave Mary the books so English allows both theme and recipient passives and this is in contrast to what we've seen before English we've seen an example of this earlier also allows both constructions and reduced relatives so we can have the books given to the woman or the woman given the books straight forward now English doesn't really might not tell us that much more but Khanti uses some interesting other data or actually shows some parallels but also shows object agreement so Khanti is a Uralic language its sister language Manjsi behaves in the very same way and both of these languages are similar to English in that certain predicates have di-transitive alternations so like English allows prepositional-dative and the double object construction Khanti and Manjsi allow so-called indirective and secundative alignment these terms basically mean that one of the two arguments in a di-transitive can become accusative and the other argument is an oblique so examples of this are shown in 32A and B in 32A we have a theme object which is accusative and the recipient which is expressed as a post-positional phrase so we have something like I gave sorry I cup to Peter gave and the two verb forms show that we can either have the verb can agree with the accusative object but the verb doesn't have to agree with the accusative object both are fine now the same verb allows a different argument structure frame in 32B we can also say I now Peter the recipient is an accusative the cup accusative case and the verb now must agree with the accusative object so we have basically an alternation like I gave a cup to Peter and then something like I provided Peter with a cup basically so two ways of expressing transfer of possession using different cases on the theme and the recipient the crucial point now is that Khanti differs from Spanish and Hindi and so on in that what is accusative here really seems to be the same kind of grammatical relation because the accusative object can undergo all kinds of syntactic processes that the other object cannot so we we see in 32 that accusative objects whether they're themes or recipients can agree with the verb that's fine accusative objects can also be in a high syntactic position outside of the VP this is shown in 33A and B where we have in 33A T cup accusative preceding so where did you take the cup where we have T cup preceding the question word where or in 33B where we have this I gave the cup to Peter and again cup preceding the recipient what they also have in common both themes accusative themes and accusative recipients is that both can be passivized just like in English so in 34A and B we see a theme passive and a recipient passive respectively in 34A we have the girl was given to him girl is in the nominative he is in the dative just like in English 34B shows a recipient passive where we have he was given a knife by John and here the recipient is now nominative and the knife is oblique in locative case and turning the page to page 14 we see that as might not be surprising by now that reduced relatives are also possible in this language headed by both a theme and both a recipient and again these kind of show nicely which case frame they originate from so it's always the accusative which can sort of become the head of the reduced relative and the other argument stays in its oblique case so in 35 we have to the child given book was very expensive so the book given to the child was very expensive and we see this marked with the latiff case which is the recipient which marks the recipient in surgut hanti in 35B we have reduced relative headed by a recipient where now we have with the book given child so the child given the book and here the child is the recipient heads the reduced relative and the other internal argument has its oblique case that it has in this case frame so basically what English and hanti both show is that it's possible in principle to have a case marker that expresses or to associate a certain case marker with one type of grammatical relation and in some sense this is similar to what we see in Spanish right because we have a similar we can have the same morphological expression for both themes and recipients but these data show that these languages differ from Spanish and Hindi etc because whatever has accusative in English and hanti actually can undergo passiveization reduced relative formation etc so we see this difference between these types of languages and English sorry Spanish but there is a way of kind of forcing Spanish into this into this pattern as well because while not all languages allow different case frames with a single predicate many languages will have different predicates which allow different case frames so I've already mentioned that in English we can say something like I give the book to the woman where we have an unmarked or an accusative theme and a prepositional or dative recipient but we can also say I provide the woman with a book in which case the recipient is accusative or morphologically unmarked and the theme with a book is now expressed with an oblique prepositional phrase English is not the only language with predicates so in Spanish one example of such a verb is armar to arm to provide someone with weapons which behaves in the same way in that its recipient is now kind of accusative and its theme is expressed with a preposition 37 shows an example the government armed the army with pistols just like in English what's interesting about these verbs now is that the direct object if you want to call it that which we can call it direct object for now is now a recipient rather than a theme and it's this recipient now which takes part in differential object marking so in 38a when we have not the army but we have an animate object we get differential object marking so we get differential object marking on the recipient now again the argument which has accusative or differential object marking can be passivized so we can get 39 soldiers who were armados etc the soldiers were armed by the government we can do the other test as well in 40 we see that the recipient which is the kind of accusative recipient now can have a reduced relative so we can say soldiers armados con pistolas we can also see that the recipient's argument the recipient argument's case now changes to genitive and a nominalization to the degree that this is acceptable so we can say el armamiento del ejercito de los soldados and we can also see that the recipient can now control a depictive secondary predicate when it's introduced by a verb like this so in 42 we get el capitán armó Amaria Boracha the captain armed Mary and Mary was drunk and and in 42b we see that that meaning is not possible when we express it sort of in the dytransitive construction with give with the similar meaning so what these examples are supposed to show is that what's what's going on with differential object marking seems to be that it happens on the on the object that the verb takes as its direct object or its primary object as it's also called in Spanish sometimes so it's not about themes and recipients per se it's about which object in Spanish at least can sort of appear as an unmarked object and when it's the recipient that it appears as a primary object so when the recipient is in this accusative role then it can undergo all of these tests that direct objects otherwise can as well but we've seen that when a recipient is introduced by and it can't do these things so it looks like the argument structure of the verb also determines what's going on here and it's we can have distinct grammatical relations here so to come to an end soon I've talked about tests sort of how we can try and distinguish between direct or primary objects and secondary or indirect objects and I mentioned in the beginning that I think that these data are best analysed in a morphological way so just very briefly on the top of page 15 I'm showing how such a morphological analysis could work and the idea here is very simple so people for a long time have been thinking about decomposing case into smaller features and sort of expressing case by using sub features so for Hindi for example we could say that nominative is actually a bundle of two features which could be minus governed, minus oblique the labels that I'm using here are from the literature in principle you can use other labels here and there's also ways of making this a bit more systematic than what I've shown here but this is just a very simple way of looking at it so nominative is specified by minus governed, minus oblique for example accusative can be expressed as plus governed minus oblique so it differs from nominative in one feature and we can have a third case dative which is expressed as plus governed, plus oblique so we get these three cases which are distinguished by one feature each now the only thing on a morphological account that needs to be said in a way we have two we have two suffixes that can be spelled out we have a zero suffix which is inserted when a case is specified as minus governed and minus oblique so whenever we have nominative in the syntax the morphology sees this and spells out nothing that's fine and sort of a common way of modeling syncretism is that you just have some under-specification in one of the items that you want to spell out so in 44B we see that it would be enough to say that core is only specified for a plus governed feature and that's the one that accusative and dative have in common so whether the syntax gives an argument an accusative or a dative doesn't matter for the morphology because the morphology will see okay we have plus gov so we'll put a core in there but that's just morphology that sort of doesn't mess with what we see in syntax and this kind of idea and there's other ways of implementing this but this kind of approach can get us the sort of different syntactic representations of of direct objects with core and indirect objects with core sort of yeah different syntactic representation of those two things but the same morphological spell out so that's the idea behind how you can do syncretism so what I tried to show is that syntax sees the differences between differential object marking objects and dative objects in Spanish and Hindi and Basque and Kashmiri if if objects with if objects with differential object marking and indirect objects are syntactically identical so if they were the same grammatical relation we would expect that this grammatical relation sort of shows consistent behavior but this is not what we seem to find in these languages and I suggested that this distinct syntactic behavior can be captured by saying okay in syntax these are different things we have what we assume for other languages as well we have direct objects and indirect objects but they happen to be spelled out in the same way on this view whatever differential object marking really is it's sort of I've been treating it as an alimorph of accusative right so in in Hindi in Spanish say sometimes the direct object has accusative case which is null and sometimes it has accusative case which is ah or core respectively ah and I think that sort of some support for this idea comes from from languages in which we know that we can express themes and recipients as the same grammatical relation like English and Hanti and these languages show different behavior from Spanish, Hindi and Kashmiri and so on so I think we can sort of put those languages to one side and say that languages in which differential object marking and dative show homophony can be explained in a morphological way because we see the distinct syntactic behavior and that's it, thank you 13 I think it might be small the exam 37, when you discussed 37 you were hesitating what to call the army primary object or direct object or privileged object does that matter for you? it only matters to the degree I think that I would say that so basically I think for cross linguistic what's the word? it makes sense maybe from a terminological point of view to say direct object is a theme object that is marked identically in a transitive and monotransitive in the and the primary object is one that is marked identically in a monotransitive, a recipient that is marked identically in the that is marked like an unmarked monotransitive object so the classical definition of those terms what I'd have to say for Spanish then is that primary objects and direct objects behave alike and indirect objects and secondary objects don't so basically that could be a way of just saying that having primary or a direct object is one grammatical relation and then there's a secondary object relation which I think it's a terminological point of view I was hesitating a bit because I hadn't introduced primary objects so I wasn't sure whether it would be confusing or not so I think, yeah, it's a it's not a very deep issue I think the idea would be that primary and direct objects behave like I've got a small maybe related question what would you call non-agreing object in Hunter in such a two-way protocol? yeah that's a good question so given that given that a non-agreing object can also passivise right, or can it not then it's different from the relation that so I'm not sure whether we'd have to introduce another object so what you do in the in the book is call it object 2 it's respected to a theme patient yeah so the non-agreing object in Hunter would be could be one that is respected to a theater role maybe that sort of doesn't make it quite clear what a secondary object would then be something like an inferior basically I think you said that it's a language with differential object marking but without a demo object construction yeah thank you very much I have a couple of questions perhaps one is I suppose a bigger question as to how you or if you can under your account for how this variation arises cross linguistically your kind of structures here but also why that would or how that would become the case and then another perhaps related or perhaps in terms of the kind of more like information structure and why so you mentioned with the earthquake example I think in Spanish sort of subtlety in interpretation I think again even with the fancy one would you have like I provided Peter with a cut perhaps being slightly distinguished from Goebbins so I wondered if you could expand on so the provided translation was just that to show a similar like case frame in English rather than I give Peter the cup which doesn't have sort of where both arguments are unmarked in provided the theme has a proposition so I provided just sort of to have a parallel in the marking the so the in so as far as I remember from the literature it's the it is the information structure with which influences which which object you have basically that seems to be perfectly I mean I'm very happy with that being an explanation right I haven't really looked into what I think for English some people have tried to come up with all kinds of explanatory variables for when you use a different when you use the propositional date of construction and double object construction and there are some restrictions on the animacy of the recipient for example so you can say things like I gave I gave Mary book or I sent Mary the book but you can't say I sent London the letter things like that so these things play a role but I think that these things will probably have chosen differences across languages so yeah when it comes to when it comes to something like using the verb to arm instead of give arms to or give weapons to I can imagine that that's also something where you know information structure influences whether you say whether you kind of want to keep continuity with the recipients or whether you in a context where you're more it's more about you know different kinds of weapons or whatever use the weapons as the theme so I think that whatever seems to be maybe more salient more topical I could imagine that there's a trend to expressing that as more logically unmarked or accused of basically with the with the earthquake example that that's yeah that's that's quite an incident I didn't look too much into I have so with Hindi the so Mohanani writes that some speakers so some writers of Hindi seem to allow this this alternation of retaining accusative and Rajesh Bhatt told me that he thinks that sort of all speakers have this which is a contradiction so I'm not quite sure what's going on there I think it's about the right is maybe I do reference the Sahar there which seems to have a similar thing that in accusatives it can and passives it can retain accusative and there it's also said that that's something a gentle about it so I haven't really looked into where that's coming from but I think it's it's about the something about the representation of the other instructor in the passive so then passive is mostly about reducing transitivity rather than sort of removing the agent and that gives us a sense of this was that whether thanks for great tool I never really know if this is like an appropriate question for these kind of talks but you've drawn on a number of languages I was just wondering why you chose these words in particular and how generalizable you think this phenomenon is it's a thing that you have to look at for each language in a way because because because yeah because we find similar overlaps right so English gives you a way of expressing themes and recipients with the same with homophonous marking in a way so the Spanish but when you look at different tests they show different results so you kind of have to have to do this I think for different languages English is handy because it's it's widely spoken by this audience for example and and gives this and has this nice transitive alternation that is well known Humpty is in there because that's sort of where I found a lot of this the similar behavior of recipients and themes and that's some accessible literature Spanish is one of the prime examples of a language of differential object marking and in general some of the papers that I referred to in the beginning where people have suggested that that differential objects are maybe expressed the same syntactic category they were dealing with Spanish in part and other Romance languages so so that's it I would like to have some more non-Indo-European languages with the same overlap and I'm pretty sure that they're around so Basque is an example but with Basque it's sort of difficult to rule out that Spanish could have had an influence on that so yeah I'd sort of I'd like some more examples which which are unrelated to the ones I have ideally if someone knows of a language like that what would be a good contrast is another language which doesn't have differential object marking but has a state of accusative syncretism because that would also show that you can have syncretism of these two cases which I'm sure also exists it's just not a part of differential object marking because that would give us another piece of evidence that there's nothing inherent about that kind of homophony that is linked to differential object marking learning languages where recipients have to be mapped as direct objects but not differential objects can give you references later so it was practicality to some degree Michael Silverstein published a paper many many years ago which he draws a linkage between what he calls normatives and datives essentially arguing from a very wide range of languages that when you get constructions in which differential object marking is collapses the encoding is always in a dative case structure so if a language for example has differential object marking in main clauses but then in dependent clauses it doesn't make that distinction then the encoding is typically dative okay of the subject then of the object then do you know which paper of his that is I'll have a look cool yeah thank you I wasn't aware of that it was actually written in about 1974 but it was published in the 90s in a volume edited by Bob Van Beylen thank you and you get the same thing in antipassive constructions so languages that have differential object marking in basic transitive clauses when you form antipassives they typically encode they collapse the distinction and encode all what would be objects as datives that's interesting yeah I'll have a look thank you which doesn't that's not what you get with indirect objects yeah I think I have to look at those structures to understand what's going on there thank you my earlier question was this how it also links to which variations yeah which variations so why why English has different and other languages don't or the other way around it's a good question so I think what some people who have looked at give a lot probably my husband has said is that I remember this correct certain kind of more prototypical and sort of give is the most prototypical one I think and I seem to have a vague recollection that if you have these kind of alternations then they would start with a verb like give so you maybe you don't have a language which only has this for send or like throw but not give so this is also something that people have looked at in English as well another language is how give and send and throw behave with respect to whether they allow these alternations yeah so maybe it's just it's a kind of extension of often information structural device from one predicate to another and it starts somewhere and it seems to start with give and it can go somewhere and then it could spread out so I think there seem to be some scales in both yeah so in this 2010 book by Maltschikov, hospital modern comery on die transitive constructions and they have this analysis I think of in terms of semantic maps where they show how how certain meanings in die transitive can spread out from give to other predicates and they show them different languages so there seems to be some regularity in how these things happen which I can't remember but yeah there is something there but give is useful because it's very common and it's easy to handle in very languages can I can I ask after the semantics in the Spanish case what is differentially marked as anemones yeah that's a that's a that's a difficult question so animosity plays a role but you can also have differentially marked inanimates I think then they have to be definite and having looked at some of the literature I mean I think it's difficult to look at all the literature on Spanish differential object marking no two people completely agree on what triggers I think one part is also one aspect of this is also that different varieties of Spanish handle these things in different ways so in general animacy and definiteness animacy and specificity play a role but there's some ways of overriding it so yeah do you know what other languages I mean is that what does the semantic factors in that's actually yeah that's that's a good point so when I started thinking about this I was actually more on the side of of thinking that maybe they maybe the two cause and so on are actually the same thing because these languages seem to involve these languages seem to involve both animacy and specificity to some degree and recipients are often also animate right but it can't be that simple because we've seen that in two we have an indefinite we have a definite but inanimate recipient or goal argument in Spanish which also takes a part so these are maximally tendencies so in but animacy might play a role in many languages in Persian which has differential object marking where the differential object marker originates from what used to be a date of suffix if I remember correctly their animacy doesn't play a role so it's also not that this date of connection is necessarily tied to animacy it just happens to be the case in many languages I think but yeah actually another good language to find would be one which has differential object marking where animacy doesn't play any role it still has the homophony because that would sort of still has differential object marking and the homophony between data because then that eliminates another variable so would give us another example of what's going on so now a bunch of go for it sometimes people compare this with object marking optionality of the Victorian so I really there's an object marking paradigm and you have to double the object if the object is animate but then there's variation across bunch of languages so this looks like a semantic plan and that's fairly robust I think and you have similar effects where it seems to be morphological terrain so in Makuwa that seems to be a noun class and then in Chaga it seems to be pro nouns so whether that's a morphological thing or a pragmatic thing a lot of knows but it seems to me that in this in a sense it goes what you say about the tendencies I think it looks a bit like you know people want to mark a particular I think semantic relation probably but it's a bit vague but then they piggyback on formal means which are already there so that's why you get the data in a sense it doesn't really matter I mean it's a useful thing that data is also a recipients of an animate but actually that might not be but it's just the formal means are there and there's a need or a desire to express a particular semantic relation so in this case you piggyback on the dated marking in the Bantu case you just piggyback on the object doubling because there isn't really a case this tension so we don't have that but from that perspective the system is actually identical it doesn't matter what the formal means is what is interesting is the sort of secondary use of established grammatic expressions and then if you think of a grammatization story you would at some such expect that that's not the primary and then you get things like what you have in German where you have remnant type things like certain predicates take particular cases like what is held with dative and think of as a genitive which are really weird but in a sense these are historical remnant for something when it was making more motivation yeah thank you that's another good point the piggybacking story actually and what I haven't really mentioned is why it would be the dative and the differential marker that would be secretive right so there's a paper by Guillape Boson who coined the term differential object marking and he wrote a lot about this and he made some counts I think and he said that when there is some syncretism of the differential marker with another case it tends to be the dative is the most frequent alternative so we can actually ask why that would be and I think that what might actually help is syncretism accounts in some senses that dative and accusative are sort of close together on people's case hierarchies so it doesn't have to be the animus in a series it doesn't have to be a formal feature that sort of gives that closeness but it could just be that something about how cases are accusative being the non-oblig case and dative being on the boundary between the non-oblig and oblique so syncretism between accusative and dative seems to be a costumistic common thing anyway so it's that kind of piggybacking but it doesn't have to be the same animus or something like that I can also add the story we've had in the book is basically that in many cases this differential object marking starts as marking of topics well we call them secondary topics but anyway they tend to be recipients for various independent in many cases that is actually the historical source of these projects then the multiplication are different ways and you know they can spread to objects which are not recipients sometimes it spreads to any object specific objects and there are different types basically we had some kind of story which goes back to the mesh stretch marking I don't know if it's true or not but that was the idea the whole story because we get differential subject marking with often with datives as well self-agent experience with subjects with datives so you get that a lot so at least in some languages when you have dative subjects you can sort of identify a thematic distinction in the predicate so that it's not an agent that would get a dative but an experience or something so speculating it's possible that language chooses to sort of build on that analogy maybe and sort of come up with something like this that you have that distinction at some point where which would be sort of more what you just said that recipients are maybe the first element that shows something and then based on that you go down a certain path you can imagine that if you have a language which marks experience some subjects with datives then that can influence grammaticalization of differential subject marking which maybe loses the distinction or loses the sensitivity to experiences but retains the dative marking or something so that's true though that then the syncretism doesn't quite work out as nicely because that close cases that's a good point there is a lot of work by Miriam about grammaticalization of differential subject marking and he'll do it okay thank you very much then I'll just