 OK, welwch yn ymwneud, mae'n gweithio'n gweithio'n gweithio'n gweithio'r Marys Cymulleru. Marys yw'n gweithio'n gweithio'r postdoc ar y Ffianna Unedig, a Marys yw'n gweithio'n gweithio'n gweithio'n Gweithio'n Gweithio, mae'r specialitys yw Maltese Linguistics, ac mae'n gweithio'n Essex, mae'r Mres, ac mae'r Universtif Sari, mae'r PHD, ac mae'n... Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! Ychydig! eich gwelwch yn rhan i ymweld i'r Llywodraeth i'r amser yn ymweld i'r llwyddiad arweithio'i'r adeilad arweithio'r Llywodraeth, iherwydd, gweld am ystod o'r meddwl i'r lleol, ac yn ystod oherwydd yn gweithio'r adeilad arweithio'r Llywodraeth. Yn hyn, ynghylch yn cyflPsuach i'r adeilad arweithio'r adeilad arweithio'r adeiladau arweithio, ydy bod y bydd y pethau'r adeilad i'w Llywodraeth, ac y rhanol i'r adeilad i'r adeilad i'r Llywodraeth. yng nghymru a'r hyn yn ei gwneud. Ydw i'r bwysig yma, Corenic Arabic Eysig, os ydych i'r ddiwrs o rhaid i'r ddweud o bwysig ddeilig, ac i'r ddiwrs o'r ddiwrs dweud o'r ddiwrs o'r ddeilig, fel yna i'r ddiwrs o'r ddiwrs o'r ddiwrs, yn cymaint o'r ddiwrs o'r ddiwrs o'r ddiwrs o'r ddiwrs, yn ei gwneud. Ite bros caught at this rate, however the same two prepositions within the same text that is within the same period can also be used as possessive markers or at least through them you can construct a possessive structure So, as you can see from 3a and 3b, essentially, the theme so possessive structure input paPERs as well as the theme argument and we observe that we've got an alternate use eich eich rhai gwahanol yn��iaid yw'r cysylltu sydd o'r rhai o gyhoeddwyr gyda'r ddysgu y gallwn sig ar gyfer gwahanol. Ower, ydych chi'n ohono'r cyfliadau cyfliad, fel ydych chi'n gwahanol iawn ar gyfer gwahanol iawn a i ffiydyll gwahanol. felly dyfodol yr ysgrifio llyfr Gruff, galloddoddoddoddoddoddoddoddoddoddoddoddoddoddodd. Dydyn ni'n yn dangos, felly y bod ni'n byddiganu wedi eu llyfr sy'n cyfreciedd traddill ym提eddoch, gyda'r cwylwyr cyfwyr deolfenol ganweithio. Nid yw bod'n dweud y fath oedd dweud llawer yn y context yn ei fath. Yn y cash iawn, dywed i'n ystyried i ystyried i'r tyntas y ddiogelion yma, y tenwch y marchaf, oedd y thymau, a ddim yi'r fath yw hynny. Rwy'n credu llyfeddau hynny'r oesaf cyfrindiau cyfrindiau, oesaf y 1997 oedd. Mae gynhyrchu bod y syniad o ymddangos inna i'r llyfau are instances of what he would refer to as a locative and goal schemas which allow us to get to a possessive interpretation with them. So this is the schema, so the locative one is Y is at X's place such that that locative construction as it were yields possessive interpretations such that X has or owns Y. Similarly the goal schema is just a change in the schema having a 2X as opposed to X's place which yields a possessive reading. Right, however Hynas account is not simply diachronic rather he through such an organization of the locatives into becoming locative schema becoming a possessive construction he also makes a clear statement with respect to the syntax of such constructions. So he believes that possessive structures which are derived out of locative and goal schemas apart from other sort of schemas are intransitive in nature such that the possessive is encoded as the subject with the possessive then being treated as a locative or goal complement. Right, so in Hynas typology of possessive structures the only transitive possessive structure is the one of the English type which is derived out of what he refers to as an action schema where diachronically it was itself already a transitive structure such that X takes Y for example that yields X has owns Y. So this is something to keep in mind that there is this interesting according to Hynas possessives derived out of locative or goal schemas are intransitive while the English type are transitive which are not of the locative and goal schemas. Okay so for Arabic additional proof that we are actually dealing with a theme that is in a subject position is not merely the agreement on the auxiliary but also the nominative marking on the theme as opposed to the genitive marking of the possessive which is embedded under the proposition. So six would be a topic-like possessive structure of classical Arabic so you see that Zaid takes a topic position whereas Karnat is your auxiliary which is yielding this tense feature. Inda is our predicate which is a proposition and then you've got the result of pronoun binding with the topic and then you've got this subject which is the theme. Okay and then you find agreement between the theme and the auxiliary and that is an intransitive structure indeed. So then when we consider locatives and possessive emotes we find that yes we're also using the same sort of preposition right and so seven is a locative construction ectip and paulu as opposed to eight which is a possessive structure paulu and ectip right. So what I really want to say is that this preposition, this and preposition has actually developed as a verb in possessive constructions and moties. Okay so while diocronically the possessive construction involves prepositions I believe okay diocronically just as it was the case in classical Arabic say. Synchronically however I will argue for a possessive construction that involves transdiverdible predicates as opposed to a possessive structure. So that would be one of the contributions in the ongoing discussion on possessive structures and moties. Additionally I want to say that the two different prepositions which are used in classical Arabic in free variation are actually allomorphically related in moties. So that would be another point that would be mentioned. Right so this is what I believe to have been the path as it were, the path of gradual change okay. So let me explain it in a bit of detail. Right so this part is what you have in classical Arabic okay. So you've got a preposition which is either followed by a pronoun or as the possessor is expressed as a noun phrase no problem and then you've got the theme as an MP right. So that was kind of the precursor of the possessive construction which actually has the same structure as a locative construction okay. Then that preceded a change which involved topicalisation such that the possessor now takes on a topic function and that topic requires binding via resultive pronoun and that is the resultive pronoun you get attached on the preposition. And the theme remains in situ in its usual position. However this is what I'll be arguing for for moties is that this middle phase has yielded this current synchronic phase whereby what we've got is a topic that has been reinterpreted as the subject. So the possessor now is the subject and then you've got the predicate which was once a preposition has now become a verb. And what was previously resultive pronoun is now actually agreement. And then you've got the noun phrase the theme noun phrase okay. And the possessor becomes optional such that you can drop it because moties like Arabic is a pro drop language. And then what you end up with is either agreement or a pronominal form that takes on a subject status depending on whether you've got the MP possessor overt or not. So that is what I believe to have been the trajectory of change and the last one is the final synchronic state. So nine would be representative of the exact equivalent to the classical Arabic structure where the possessor is interpreted as a topic which is bound to the argument of and the preposition that is. And the theme is the subject so that is exactly the same structure which we had for Arabic okay which is for classical Arabic which is the synchronic state of affairs for classical Arabic. What I want to say is that this is not the synchronic state of affairs moties but rather this is such that the possessive construction is transitive in nature where Paulu is actually our subject agreement and that's a verb now. And tip takes on an object function okay. So this is what this is the end state okay. But now I'm going to give you the evidence why I believe or I'm assuming that this is truly the end state okay. So I'm going to give evidence both for the verbal nature of and in this context as opposed to the prepositional nature which was its kind of precursor. As well as more evidence that the possessor is really a subject of moties and the theme is really an object of moties as opposed to the other or their other mapping which we've got in classical Arabic. Okay so this will be the set of evidence I'll be considering essentially where the other negation realisation differences. The behaviour of QP subjects agreement and accusative marking of the team okay. Right so in general. If you've got locative so blue will be my locatives okay and green will be my possessives and hence one will be a P and one will be a verb okay. So in 11 you've got an order such that the team is preceding a prepositional phrase okay. So and is functioning and in blue which is locative therefore is functioning in the same way just as any other proposition and the theme is preceding okay. Yet when you consider the possessive construction which I believe to be a verb we observe that themes typically follow verbs when it comes to verbs. So you've got tip which is a theme which is following a usual verb form and in this context in the possession we observe that the theme is actually following the possessor. Another word order effect is that if we were to shuffle around the constituents in a locative structure we will need the insertion of the existential profile which is M moties okay. So 13A represents a shuffled around like it's not the canonical order so we can displace things around. So the displacement of made from its usual canonical position requires the presence of M the existential M. Removing it will yield to ungrammaticality. On the other hand it will still believe that made up is the theme which should be equivalent to made up the theme in the locative structure. Then nothing would account for the fact that we do not need to have the M insertion in this context in the possessive. Of course M is possible in the realm of and right. However the only reading you can ever get with M is a locative one. So 15A is simply a normal locative structure, non-shuffled that is in its canonical word order. So etyplach marant etyfel. However shuffling around the constituents or presenting more emphasis say M etyplach marant etyfel would only yield the locative use, the locative interpretation of and okay. So that's one difference. Then negation realisation. I think this is one of the most robust tests. So a finite negation in Maltese involves this use of two parts as it were. Ammar and the shuffling attached as opposed to negation of a preposition for example, a preposition and predicate. In which case you'd need to get like a pronominal form which includes the Marcia so we'll just refer to that as pronominal negation. But I hope you can see that there is a difference between the predicate is there and then you've got Marcia attached to it. The predicate is full and then you get a pronominal negation which is negating the whole constituent. So then if we insert and we observe that and as a verb so that is and in its possessive use actually yields the Marcia attachment just as verbs. Whereas the negation of the locative use of and requires us to use a pronominal form. Okay. So moosh or me shandom. Right. So another test. So for Maltese it has been observed that while quantifier phases can actually be subjects. So like in 18 this is an NP Paolu and cwltifel is every boy is a QP quantifier phase that can be a subject. Fine. Yet QPs as topics that are bound to result a pronoun. So they form they are part of a clitic left dislocation structure are not possible as topics. So while Paolu as an NP is okay to be bound by the pronoun on Ra on the verb a QP is not possible. Okay. So I can't have a cwltifel right to my right to Maria. However in the case of our possessive structure it's possible to get cwltifel and lepdotifel which are both QPs. So if we were to say that the OO on and is still the result of pronoun attached onto a preposition then these can't be topics themselves. That means therefore that these must be subjects and the OO on and must be analyzed as an agreement market as opposed to an entrepreneur attached onto a P. So I think these are all pushing towards the verbal use of and. Right. Another fact that's agreement with the auxiliary kin. So what we observe is that the use of and in this possessive use allow us to get either default agreement on the past tense auxiliary or as agreement with the possessive. Okay. As opposed to what we get in classical Arabic which is agreement with the theme. Right. So that's count one is G which is coindexed with the E and D which is also one is G. So this is different however from the obligatory agreement which you get on the same auxiliary in a possessive in a locative structure. Okay. So agreement in this context must be with the real subject which is the theme in the locative construction. Okay. El codba. And then you have to have kinu because codba is plural and hence you get 3PL agreement. So all this is another difference between the two. A further difference is the agreement facts on raising which I on raising predicates which I believe is another robust proof that we're dealing with a possessor that has become a subject of Maltese. So if we take the a specializer regia to repeat to do again we've we observe that it's a it's an obligatory subject subject raising predicate. So row. So this is it so happens that Maltese does not have no finite forms really. So don't bother about the fact that we don't have the two equivalent in English. Okay. It's finite raising is possible in Maltese. So in the embedded clause you've got a third person inflected third person plural inflected ver form. And then you get the same agreement by raising on the on the a specializer. So getting a default reading is yields to one grammaticality. Okay. So that implies that this is a subject subject raising predicate. Right. So then if you insert and the possessive and in the embedded clause you find that you actually need to have agreement chain. Does it work right? So the third person feminine agreement on and which is representative of the possessor actually becomes the agreement. On regia, which is which itself requires subject subject raising. Okay. So you get three sgf agreement there. Same follows for spittrow, which is another specializer. And then the same facts follow for the scene predicate. Okay. Which is the canonical raising predicate as well. Right. It's true however that in the matrix we could have had a third person singular masculine form in this context. But I argue that that would be the equivalent of the it expletive in the equivalent of it expletive in English. Like it seems that as opposed to she seems to have gone. Okay. Right. So now so the evidence which we had so far was both for the subject status of the possessor as well as the verb like status of the predicate as opposed to its original prepositional status. Now what I'm going to argue for is the grammatical function of the theme. Remember that the theme is the subject in classical Arabic. Whereas what I'm arguing for in Maltese and extends for the dialects really is that the theme is really an object. Right. So Maltese happens to be blessed with differential object marking, which I am really relying on to establish family that we're actually dealing with an object team. So depending on how high up on the intimacy hierarchy you are, Maltese displays little marking or non-lil marking on the object as opposed to the second object. Second objects are always marked. Okay. So this is why it's a differential object marking as opposed to object to or secondary object marking. Right. So this is a transitive verb C. So what we see is that when you've got girl, which is human yet indefinite, little marking is impossible. Okay. You've got to say right. However, when you get definiteness on that same argument, little marking becomes optional. However, when you go even higher on the intimacy hierarchy such that you get you get a proper noun, which is the maximum you can go terms of referentiality and intimacy. Then you have to have the little marking. Okay. So this is what you get with a canonical verb predicate. And lo and behold, this is exactly the same pattern you get with and. Okay. So depending on whether you possessive and that is. So depending on what the team looks like, as it were, you get the appropriate little marking or not. So I take this to really imply to me that the theme in Maltese is really an object. Moreover, another fact is that if we consider the use of and in blue now, hence in its locative use, which also takes an object, right? Because prepositions take objects as their argument. We observe that prepositional uses do not little mark their object. So this is, so little marking is solely something related to verbal predicates. So it's just verbal predicates that mark their object with little. Okay. So in this context and in the locative use is functioning just as any other preposition in the language. Okay. Such that it is not little marking its object argument. Good. Okay. So to conclude this little summary here. Basically I think that what we've got here is a change via morphological conversion. So there is really nothing which is yielding the change as it were from a minor category. That's a proposition to a major class category, which is a verb, which is a bit non prototypical. Such a change. And I believe that syntactically the change must have taken place via a reanalysis of the structures topic into a subject. And Cymru has an argument for this in saying that yes indeed there is a correlation between the fact that the possessor is high in anonymity usually. And that would relate it easily to a topic. And similarly there is another correlation between topic hood and subject hood. So it's easy for such a change to have taken place via a topolised structure. And then consequently if we initially went through a topolisation stage and then we ended up with a transdiff structure. Then yes there must have been an important reanalysis of the resultive pronoun into an agreement marker. And then eventually the optional dropping of the possessory noun phrase itself. Yes. So all this, the grammaticalisation of the pronoun especially has resulted in the formation of this class of predicates which are referred to as pseudo verbs in the literature of Maltese and Arabic. So essentially these are referred to as pseudo verbs because while they take verbal uses and auxiliary interpretations sometimes as well. And features which are typically associated with auxiliaries or verbs. Nonetheless the stem form is not verbal really, just as und. So we know that und came out of that preposition. So that makes them kind of a class of their own. Additionally the subject ends up being expressed non canonically as well. So instead of the usual nominist of inflection you end up with a genitive or accusative inflection of form. So if we consider this paradigm, if we're saying now that und our possessive use is actually a verb just like any other. The difference however is that it's a completely different paradigm. So this would be the usual paradigm of a canonical verb form in Maltese which would be split on the basis of perfective and perfective. And that's nominative inflection. Whereas on the other hand we're saying that und is functioning as a verb in possessive constructions at least. Hence the E, Ech etc. is really realizing the subject. And for that matter therefore this is a non canonically realized subject. And this happens to be genitive inflection. It could have been accusative inflection. So far I've just discussed und to be honest. But law has to come in at some point because I had mentioned that classical Arabic has two prepositions. And this is a journey which looks at both und and law. What I have to mention prior to this is that all that I've said for und follows exactly for law. Such that historically it was a preposition but synchronically it must be functioning as a verb. The further complexity which law involves is that it is never found at least as a verb form. It's never found hanging on its own as it were. It's never found like und. Rather it's always part of a larger word form. So look at this le here within the word form kellw. And look at this le there which is part of the word form ikollw. So I want you to keep in mind this split. Undu on one hand and kellw ikollw and the rest of the inflections on the other. Right. So this is what I believe to be the trajectory of change that took place in Maltese. So this part is fairly uncontroversial. So we started off with some sort of preposition which is law. And in classical Arabic it had both a possessive use and the locative use. And from that we know, because we have it synchronically in the language, that the preposition in law is still a preposition in law in some instances in Maltese. Additionally it also developed as a case marker. And that case marker was initially a dative and then it shifted on to becoming an accusative. Yet this part now is what I wish to share with you. So the red part of the diagram here is something which we do not have evidence for synchronically. We don't have no evidence to claim data that is to claim that law was ever a verb form that expressed a possessive relationship in Maltese. However what I'll be trying to do in the range of the story is to actually get to that inference via what goes on elsewhere in the language. Then and comes in at some point in the history. It was once a preposition. Then it itself changed into becoming a verb. And somehow, and this is what I want to show you, somehow it fused in the paradigm of law. They were both being used as possessive predicates as they are synchronically in the language. And then from then on they also shifted off to develop further that is into becoming modal auxiliaries as well. Right, so let's go through that journey. Okay, so law is a preposition. That's no problem in the sense that the two there is in front of a locative. In C there is true that God is kind of a human being but the interpretation of this is literally from lip to God is like God as a destination. So I'm glossing it as a preposition in this context. So that's the use of law as a preposition with allative case possibly as well. Then law as a case marker. I think it's fairly established as well that something like I didn't send the letter to you. That should have been the letter to you, not the letter. So letter is the direct object which is co-refer to through this accusative pronominal form which is 3SGF because letter is singular feminine and that is marking the second person which is the indirect object. That's what I meant to say, indirect object. So to you. Okay, I sent the letter to you. Then it has also developed as I had already shown as an accusative marker. So Paul which is here, structurally it is the topic yet it is bound to the object of C of the verb to C and hence because it's an object it has to take accusative marking and that's the le which you see on Paulu there. Right. So then if I'm somehow saying that synchronically 36 is ungrammatical. Okay, something like Paulu Lahu or Lou Ktip is impossible in Maltese. Then this seems to be in conflict, right, with what I'm trying to say that once upon a time La was indeed a possessive predicate in Maltese. However, what I wish to share with you is that I think there is two pieces at least of evidence which we can use to infer that once upon a time in old Maltese that was in fact possible. Okay, right. So let's consider what's been said on Maltese in the literature. So every discussion on Maltese possessives always illustrate the ungrammaticality of 38. So 38 is really ungrammatical. So it's impossible to get and the possessive and in the context of an auxiliary to be. Okay. So the claim therefore, the inference, the common inference, is that kell and the call, which I told you to keep in mind, that paradigm which we saw earlier, the claim is that in non-present tense context and cantocer, cantapier, in which case the language needs to resort to the kell form and the call form, which the literature treats as being in a paradigmatic relationship. So what I want to say, however, is that rather than being in a paradigmatic relationship, kell, la and and are in an allomorphic relationship. So what I wish to push here is that la and and are actually on the same level, even synchronically. So this is how I get to and I'm going to tell you the evidence for that now. So this would be the structure. I am assuming for something like Paul had a book. Remember that had is a context where it's a non-present tense context and hence and should not be able to okay. Given the ungrammaticality of kin and. Right. So I'm using at the back of my analysis there's lexical functional grammar, which is very surfacing, a base theory. But essentially what this is showing simply is that this is a normal subject. Okay. That's the possessor. And then keep the team is an internal argument to the to the verb to the predicate, which is have. What I'm saying further is that kellw is really the fusion of the auxiliary, which is the auxiliary kin, the auxiliary to be, which is yielding simply tense features and the diachronic preposition la, which has now become a verb form. And together. They end up with the lexical form kellw. And actually this is what's referred to as lexical sharing. This is exactly the analysis you get for. I'll go in. I'll go like I will go in English. Okay. Where you've got the fusion, literally the fusion of the subject along with the auxiliary. So that's how LFG lexical functional grammar would treat. The fact that you've got literally one lexical item, but with two completely different functions. Okay. That would be the fusion of the end node and the V node, whereas what we've got in the I node, whereas what we've got in Maltese is the fusion of the I node and the V node. So how do I get to that? I think the major evidence which I am using in order to claim that that is the case, that that is really the analysis, is because it's due to something that has not been discussed in the literature prior to my dissertation as to it. So while it has been said that and never occurs in, in non-present tense context. Well, this data actually shows you that yes indeed, you can have and in context of non-present tense. Because kin is what's providing you with a past tense. And yeah, this is also kin, which is past tense. Okay. So this date exists. Okay. So it's not the case that and cannot be used in the context of kin, which is the past tense. Right. So this is, this is literally the analysis of the morphology, the morphology and the syntax in one slide, to be honest. So what seems to be going on, I think is that the possessive predicate. Okay. So yeah. Sorry. The crucial for me is the presence of something which is coming in between kin and and. Okay. That is crucial for me because if I remove them, if I remove this already and that's still, I end up with an ungrammatical sentence. Okay. Just as we had earlier, we had an ungrammatical sentence earlier. Right. So what I'm saying is that the possessive predicate and what does actually exhibit is simply an adjacency restriction whereby the adjacency of the possessive predicate with the auxiliary, kin in both its perfective and imperfective forms, actually comes to trigger the presence of a suppletive alamorphic form. Okay. And that's how you get the fusion of kin and la. Since you can't have kin and and together and changes its form as it were to la and you end up with kin la and via morphophonological result of enclitizatio, because la is really prone to erosion in Maltese, you end up with this enclitizatio of the predicate on to the auxiliary and morphosyntactically then you end up with lexical sharing. So that's how I arrived to the alamorphic relationship being simply triggered by this adjacency restriction apparently. Right. So at this point where are we? Okay. It's true that there is no actual data to be honest that I can claim that la was synchronic, was there chronically really a possessive predicate in Maltese. Yet I believe that we can still get to the fact that from this we can still infer that la was present in old Maltese. So how? So at this point we observe that the original preposition and must have developed as a verb. Okay. And at some stage, as a verb form, it has fused into the paradigm of la. So that implies that la could have existed as the only verbal possessive predicate in Maltese, with and possibly being used also as a possessive predicate, but possibly with preposition and status, who knows, if it was even present as a possessive predicate. So when and comes in as a verb, it seems to wipe off the older use of la. Yet it comes to demonstrate this sub-jacency constraint, which is by the way not demonstrated by the prepositional counterpart, because we can have ilk typ kin and paulu. Okay. So kin and in the locative construction is a possible combination, as opposed to kin and in the possessive construction, which is ungrammetical. Okay. So there is a difference there. So apparently this adjacency constraint at least at the verbal use of and seems to block the use of and. And then the pred form la comes to be used. Then what happens is that ankleicisation of la at some point took place. So I'm assuming that possibly kin la was available. And then you end up with erosion of the la. And then ankleicisation takes place such that you end up with the form kell. Okay. And what's really, really important for my statement that it was already a verb form is that if you observe the incorporation, let me just give you an example to see the gloss especially. Sorry. Yeah. So if you consider the way I'm glossing kell and equal, you might have observed that this is actually third person. This is imperfective, singular masculine. And this is B, perfective, third person, singular masculine. So with that, this is the only possible verb form, okay, with respect to the auxiliary. So with that, I assume that since you had already undergone, since you were already observing this change, it implies that the grammatical function, thematic roles had already been switched, had already developed from what we had in classical Arabic. Because remember, in classical Arabic, the auxiliary actually has to agree with the theme. And the theme need not always be a third person singular masculine, could have been feminine, right? Yet the re-ordering, the re-mapping which you observe then once the possessive structure changes from an intransitive one into a transitive one, and then the switch in the grammatical function and theta role mapping such that the theme is no longer the subject but is now the object, ends up with the auxiliary being defaulting to the third person singular masculine, okay? So the fact that the incorporation of the predicate law onto the auxiliary is onto a third person singular masculine auxiliary, I take to imply as being indicative of the mapping which had already taken place. So my belief that this allomorphic relationship seems to have become at some point so robust that it really shifted off together, developed further, like in a block. So it's not that and went on its own and law went on its own, rather as a block in an allomorphic relationship, they shifted into also becoming modal auxiliaries. And this shift, however, is not present in the rest of the Arabic vernaculars. However, I should just mention that in Latakian-Syrian, the construction Ili Alp, which is literally I have a heart, okay, which is a pure possessive construction, has now come to be able to subcategorise for a clause, okay? And together with a following clause it comes to yield a half two interpretation, so a modal use, okay? So this is the closest example I could find from the other vernaculars where a possessive structure is kind of developing or yielding modal value. So what I want to say is that the allomorphically related possessive predicates and and law have also developed as modal auxiliaries and moties. And this is actually mentioned, the modal use is mentioned in the literature. What they don't do, however, is the connection between the possessive predicate and the modal predicate, which is, by the way, as I will show you, a common trajectory of grammaticalisation across languages, okay? So these are different uses of the and and the call and the call as modal auxiliaries. So what's beautiful about 43 is that it uses and in this context, the first and is the real modal use, and equal, the second that follows, is actually the possessive use. So this is a free relative, actually, free relative clause. So minias elitre ittaiba, whoever chooses the good way, and we call it prosperity. So have, the modal have, has to have, as it were, has to have. So that is the modal have, this is the possessive have, okay? And that's the theme of the possessive. So that really goes to show the robustness of the grammaticalisation of and and the rest of the paradigm as modal auxiliaries, because they can then take a possessive predicate. Right. What's crucial, however, to mention is that the claim is that the fusion which took place, apart from the fact that they refer to them as being in a paradigmatic relationship, which I'm saying that they are actually in an allomorphic relationship and not a paradigmatic one. The additional fact that all the previous literature says is that these irregular forms are actually derived out of the combination of the auxiliary with the preposition. Whereas what I have been trying to show you all along is that this is actually the verb form that we're dealing with and no longer the old auxiliary. So this is not the case that we're dealing with an auxiliary. Rather, we're dealing with a morph form, it's true, Calendicol, but that morph form does not involve that preposition but actually a verb form. And I believe that to have demonstrated that the possessive predicate and and Maltese is actually a V and not a preposition and to have just posited an allomorphic relationship, then it should follow that La should also be of the same category. You can't have an allomorphic relationship between a verb form and a preposition. They should at least share a category. Right. So in my dissertation, however, I provide an alternative analysis, the possibility of treating the le part as a date of marker. However, now I scrap it because that can't be either. So the possibilities are that it's a preposition or a date of, since as I showed you it has developed as a date of marker in the language or as a V. So now I'm positing a V analysis, but in my dissertation I had posited a date of analysis. Whereas that can't be why because if we consider a proper date of use, so the recipient, so the le here, so as for the children, she sent the book to them yesterday. So this is the to them, is it right? This is introducing the to them. Doubling that pronoun requires the presence of a date of marked NP or as a date of marked pronominal form, but not a nominative. Okay. So I can't have, sorry, that should be date. This is, sorry, this is a typo there. Actually this is a typo as well. Okay. So what I meant to show is that the doubling of this pronoun, which is attached on to the verb, must be via a date of marked NP or pronom. Okay. So litfal a lidom. However, the doubling of the le followed by its argument, in the case of cal and equal, must actually be not via the date, but via the nominative. Okay. So I think that goes to show that it can't be the case that that is a date of marker. Because if it were the case then we would expect doubling to be via the date of marker as well. But that's not the case. However, this is the last point essentially. There is another piece of proof to help me further deduce that we are really dealing with a law that was already a verb form as opposed to a proposition. Okay. So back to our, so this we got to now. Okay. We've shown this part as it went. Now we're still stuck with this, but I'm going to show you that because we've got this, we're going to infer that this was the case. And I'm going to make extensive use of grammaticalisation theory in this context. So what the red part is representative now of, now is what we're going to try to infer from the very presence of this additional grammaticalisation in the language. So in order for something to have become an auxiliary that meant that at least there was something verbal preceding it. So we're still stuck with 48, which is a grammatical. And it's like we're trying to say that that was really possible in old multis. And I'm going to, I think it's possible to infer that that was really the case via these three well-known clines present cross-linguistically with the help of grammaticalising theory. So essentially grammatical morphemes are always canonically at least, always derived out of lexical morphemes. And in order to get to an auxiliary and even further aclytic, we have to have a full verb somewhere. Okay. And then it ends up as an ethics of some sort. Additionally, it has been shown cross-linguistically that features important morphosyntactic and morphosematic features such as perfect, resultative, tense and mood have been shown to cross-linguistically frequently been derived from possession schemas. Actually, if you take English, English is a canonical instance. Okay. So you have the verb have, which is used for both possession as well as for a model as well as for a perfect use. Okay. So in order for all that to have been derived, there must have been a common V at least verb. So this was something which had never been mentioned before, which is really nice I think that actually the possessive predicate la has indeed a motif developed as a universal perfect marker. And what's beautiful is that I find this at the station, the first one, from Cantalena. Cantalena is the oldest ever written multi-script. We don't have something earlier, unfortunately. But what's beautiful is that 1470, for sure, this structure was already grammaticalized. So basically this structure, as you can see, this is the ill form, okay, which is the la. Okay. But as I said, it's prone to erosion. What's happening here is that to me, literally this would be, this would literally be to me, illi, zmin. So to me time. So that construction is equivalent to the possessive structures we've just saw for classical Arabic, to me, a bread, for example. And in this context, this is to me time, and then you've got another embedded clause. And together that construction actually yields the universal perfect and multi. So this is how we are able to say, I've been building, for example, or we haven't been talking for a long time, for example. So in order to assume that this la is now synchronically a grammatical morphym of some sort, at least it's providing us with a perfect aspect. So it is grammatical in this respect. Then at some point the original proposition must have underwent a change into it becoming a verb form, I believe. What's beautiful is that at the same time Holman discovered that indeed in Syrian you get exactly the same construction. So this is something which classical Arabic does not have, but all the dialects have. So this is something so beautiful such that the same la, which is also found in the rest of the dialects as a possessive predicate, but unfortunately I'm saying in Maltese this was present in old Maltese but we only have fragments of it as it were. The same process took place. So there are good reasons to believe that the grammaticalisation of the possessive predicate into a perfect marker must have taken place quite early in the language and that would require that this was once a verb form for it to have become an auxiliary, a pure lexical verb form that is. So a possessive predicate which is verbal in nature. So the reason to believe that this was an early grammaticalisation is from the fact that eventually there is no alternation with and. So well you could argue there are possible, I mean multiple ways how to question this. Why is this not the case? Well possibly and was not a possessive predicate at all and was purely a locative one or else if it were also a possessive predicate it did not yet have verbal status. It was still a proposition. Secondly, if they were actually both available in the language as possessive predicates, I mean as verbs that is, in principle nothing would have restricted the law to develop as a perfect marker as opposed to and from developing as a perfect marker. Why? Because Tunisian actually which does not have the use of law as a possessive predicate has developed and which is the only possessive predicate which they have available as their universal perfect realising auxiliary. That is beautiful. I mean it could have been either way in principle if it were true that we had both and and law functioning as possessive predicates and moties. Secondly, it must have been an earlier construction. Why? Because the past tense realisation does not involve an inclinicisation. So here we observe that ilu, il, the same law is in the context of the auxiliary b but this time there is no fusion such that this could have become kellu or kellu, nothing prohibitive. Yet this is not the case. So possibly at that stage in the language's history law still had like a full phonological status as opposed to now where it is highly, highly eroded. Unfortunately that's not just it. There is an additional twist. So that's how I think we end up with a verb form, OK, so far. Yet there is an additional grammaticalisation. Yeah, if you want to call it grammaticalisation. So I've just said that in my analysis at least that kellu and ikol are ffusions of an inode an inflectional node as well as a verb phonode. Right, and they yield that one. Yet what's happening now synchronically which is something I haven't encountered in the Arabic dialects though is that it seems that kell and ikol are now becoming bleached of their, once upon a time, tense and mood value. How come? So what's happening is that you can actually repeat the auxiliary now. So you can have something like kont, so that's our typical auxiliary, kon ja kellek and then you have another kellek. So I had initially mentioned that kellek is really the fusion of i plus v. What I'm saying now is that it can't be the case any longer in these constructions because there can be only one inode, right? So, and this is the real tense barrier whereas this one is simply functioning as a perfective predicate. In fact it yields past perfect just as any other perfective predicate. Same here, so say kun, so this would be your auxiliary, irrealis or habitual depending on how it's used. And then you get kellum as well. Kin equally, right? So the evidence I use in thinking that what was once, or what's still lexical sharing at one part of the grammar can't be the case in this context, is because essentially the interpretation of the auxiliary plus kell and ikol and their inflections is really the equivalent of any other lexical predicate you could have had in the language. So the combination of, for example, kin, perfective kin, along with a perfective lexical verb form yields the past perfective motifs. And that's exactly what you get in konja kellek. So kellek, if I had any other verb, any other perfective verb, the interpretation is past perfect, okay? So that's the combination. Crucially, there is a major constraint in the language such that with respect to auxiliary ordering that the imperfective form of ikun must always follow the perfective form of kin, okay? If we still want to believe, which we're not doing, if we still want to believe that there is still some sort of I, so that is inflection interpretations in kell and ikol in such context, then it would be impossible to account for B because in B we observe ikun, but which is preceding kell. And if kell is really synchronically the fusion of I plus V, then it's as though we're saying we've got kin under ikun. So that's the perfective form of the auxiliary under the imperfective, and that wouldn't be possible, okay, to say that. So we have to assume that in this context this is no longer involving lexical sharing between an I node and a V node, but rather it ends up being a verb form in itself. Right, so to conclude, old multis must have had a process of construction, I believe, in which the predicate law was used across the board, yes, and from a preposition it had developed into a verb already. Synchronically, there are contexts where yes, a verb and auxiliary forms, la and and, are actually in an allomorphic relationship. And another, I think, beautiful illustration is that the grammaticalization of the possessive structure in multis has taken place twice, but at different times, at different points in history. So the initial grammaticalization was when la was the only possessive predicate out of which you get the universal perfect in multis. And secondly, at a later stage, that is, this had already come about, the allomorphic relationship, out of which possessive, so the possessive structure which involved this relationship, you end up getting the modal interpretations. So that was it, really. Although I can show you, is there a time what? Questions, questions, whichever. No, basically I was going to show that Hebrew is doing the same thing. Literary Hebrew is exactly like classical, yet colloquial Hebrew is doing something different, which is exactly pushing towards a transitive structure. So, as you can see, I think it's clear that you get the same la marking in Hebrew. This is literary Hebrew, and agreement is usually with the theme, with watch in that context. Yes, what's happening is that, first of all, if you get, while the possessive takes nominative case, apparently that is nominative, the theme is nominative in literary Hebrew, so the possessive, the theme that is. In colloquial Hebrew, this is taking et marking, so et marking is the accusative equivalent. So that is a change which we also saw for multis, where the theme got the little marking. Secondly, in literary Hebrew, agreement on the auxiliary that is controlled by the possessive, whereas now in colloquial Hebrew, the agreement is no longer being controlled by, agreement on the auxiliary is no longer being controlled by that. Secondly, while the possessive could undergo subject to subject raising, so it would be the watch which agreed with the matrix predicate, in colloquial Hebrew, this is no longer the case. Additionally, in terms of word order, the possessive, the possessor is shifting its position to come before the auxiliary. So that's exactly it in the case of colloquial Hebrew. So, according to Cymru, many Arabic vernaculars have undergone exactly the same changes for what we see in multis, and when it comes to the agreement on the auxiliary, because that is really telling, his claim is that the only, virtually the only option is to get terpers in singular marking on the auxiliary, as opposed to the team agreement which you observe in classical Arabic. However, the wrong conclusion that's deduced at least in Cymru in 1991 is the claim that, because of such a default 3SGM marking on the auxiliary, such constructions do not have a subject. This is his quote. This suggests that in multis and by extension, the Leventine dialects, in them, neither argument is subject. Indeed, the sentence has no subject is in person, so this is the claim in Cymru, and I've shown good evidence that actually these possessive structures actually do have a subject, and the subject is the possessor. Right, so if you then consider the dialects themselves, possessive constructions in Palestinian, well, you get 3SGM marking, even though in principle you could have had agreement with the team, yet, so something like 64 is really marginal. So agreement with the team, which is feminine, girl, is very marginal, but rather it is 3SGM that you default to a person's singular masculine that you get. What's beautiful also is that they can get the negation, which is the one with which you usually express verb forms as opposed to prepositions. However, then in Tunisian and Moroccan dialects, the situation is a bit more complex in that, literally, they can have default agreement, agreement with the team, as well as agreement with the possessor. So in whichever way you look at it, crucial is that for sure there has been a shift in the mapping between the grammatical functions and the thematic roles when considering the development from classical Arabic to the dialects reading. That's it. Good. Does anybody have any questions you'd like to ask? OK, I'll start with one question. I guess most relevant to this, exactly what you're saying here now, is I'm wondering, you know, on your story, why is it possible at all to say, kin anni? Right. I think that... Then let me explain. So, I, why it is, you're trying to say, OK, you need to have a thing in between, either Jack or whatever, kin Jack anni. But the point is, the possessive andi, which is first person singular, and the past auxiliary can agree with that in person. But it doesn't have to. Whereas with a true verb, it would have to agree in person. So how come it's allowed to not? Right. My story is simply that this is a non-canonical realised subject. So this is genitive inflection as opposed to the canonical nominative inflection. And hers goes on in other languages that I can write at all, 2001. They really show that in many places, when you get, when you get non-canonical realisations of, for example, object or subject, you simply default to whatever your language defaults to. And it happens that in Arabic, it is the default to 3SG and form. That would be my answer, to be honest. And I mean, if you look at Syrian and Maltese, agreement is allowed, but people will always go to the default form. So I could say, con John, there as well. Yet, I think if you do a survey, people would prefer the 3SG. It's like, yeah. I was wondering how the kind of stories that you present here, fit with other commercialisation that goes within a language. I'm particularly interested in, like, development of auxiliaries and murderings. So, yeah. So, models and Maltese are, and Arabic, because they're the same. Are really, are really all pseudo-webs. So pseudo-webs are exactly the sort of things which I had been discussing here. So they really are a class of their own. And this and, however, is not a model in the other dialects. So there is an argument in the literature by Van Hoff, who believes that influence from our Italian ancestors, as it were, when they colonised the island. So she believes that there is a context story for the grammaticalisation of this paradigm as a model. Yet, in general, they are really an odd class of predicates. It's true. They are a very odd class of predicates models in Arabic. Yeah, they are not the canonical verb forms at all. So they could actually be, they could have just a default tree as teripresence in feminine masculine form only, for example. The only thing which is nice, I mean, what's nice with respect to this specific model is that it follows trajectory of changes from possessive structures to other, which is something which is well known. It's just that apparently no one, at least Heiner, which I read so much, believes that this is really not something that goes beyond Indo-European, but this is actually something which we're saying goes beyond, especially the universal perfect as well, which is present in the other dialects. Yeah. You had another? Did you have another? You do another one. Okay, yeah, this was actually a smaller question. So one, you had examples 11 and 12. So right at the beginning when you were introducing so much, yeah. Yeah. So not knowing enough about the language and how it does things. Can you ever end up with context in which it's ambiguous in these kind of things? No. So not having in 12 B for example where you obviously have some marking, do you ever have context in which the language is different? There are a lot of factors playing a role. So for example, the definiteness of the theme typically in a possessive structure typically you'd get an indefinite theme as opposed to in a locative structure where the theme is typically definite. That is one. Secondly a crucial point is that while in B that O is really agreement and you can't change it with an NP the argument of the real preposition and can alter it between being a pronoun as well as an NP. So those two factors are really crucial. And while that is the canonical order of each I mean shifting them would result in a topic structure. So in this case il ctip in the prepositional one in the locative one in 11 B is actually that is its canonical position as opposed to ctip in the possessive which is in its canonical position as it were. So in that way no, there wouldn't be ambiguity in that way. You will need the progressive marker, ed, an ddo ed il ctip. I already thought about that. I would need to insert it. Paolw, an ddo e'n saf o'r ed il ctip. You need the progressive which kind of locates it further. Yes, literally at him is being booked or is located booked. Yeah. The differential object of marking. Happened with top and ends? No, not really. Because in B the definite allows optional case marking. Sorry, in B. The accusive marking is optional as opposed to c where it's obligatory. That is kind of the minute difference. In C part together correct? With an indefiniment in A you may not have the You can't, exactly. Yeah, sorry. It's just notation. Why should it be that you can't be adjacent to people? Because in other Arabic where I see this as fun. Well, in its prepositional state function it's definitely not. In the possessive function it's also fine. I mean... I mean I think that gives me a much more of a point in saying that this could be really a constraint which is what is triggering. It's like a condition. From a kind of synchronic point of view I think it's very nice. I'm interested to know why this change has come about. I mean, can you think of other anything else in the language where there isn't some kind of intervening adverb that gets unbomatic? Yeah, it's an adverb. It's like a blocking effect. You may just follow up on that point. There's obviously no difference in phrasing, intonation when you put in an adverb. I mean, that's why I chose on purpose an adverb because if it were something else then it would be a clause. I mean, you're in another clause, it's another predicate of some sort and no no change in intonation at all. Yeah. What is important however is that for now it seems to me that it's a finite class of things which can come in between. So it's really three items which I found which is the still, already and perhaps, forcing. Else, I'll just default to kell which is the That doesn't link to the kind of perfective aspect you want to add. No, at that point, no. You're still pure past tense. Exactly. So that's why at that point I believe that we're still I plus V as it were. But then when you get the kin and the skell formation and the call formation, that's clearly something which is devoid of tense features at all because the I position now is filled again with some sort. And even the interpretation it's purely perfective as opposed to past. But yes, it's odd. Yes. I don't know. Okay. And also, I don't know how you come to the conclusion that La is an accused marker even in this example. Why wouldn't... Why would it make you decide it's an accused marker? Well, I'm assuming. So what would be the alternative? Still, the position. But there's something commenting between the verb and the verb. So another interpretation would be I saw something the verb to the verb. I saw something? Yeah, I saw the verb, or I found the verb to give it to you. Ah, you mean, why is this not being analysed as a... Incorrectly, only vowels are accused markers in Arabic. And you can think of any variety of Arabic where there is a consonant or well, La is not a pure consonant but still both present as a consonant as used as a case marker. Well, I mean in the literature, La as a date of case marker is well known even for the dialects themselves. So if you take Egyptian, for example, which No, I'm talking about the dialects. So if you take Egyptian which is really neat because it has the Marsher negation just like Maltese what happens is that you can still get the le pronoun inside the verbal construct. So you get the verb form and then you get ma in front and then you get two pronouns because you've got two slots available for the accusative and the dative and then you kind of close it with the sh. And that gives you evidence that this is really a verb form which includes within it two pronominal forms, one of which is dative marked. So there are arguments in the literature that the le as a dative realising the recipient or the goal or the secondary object or the indirect object whatever you want to call it in that use the le has developed as a case marker. There is evidence for this in the literature. Now what Maltese seems to have done is to kind of take this a step further. It's true that this thing, this fact does not happen in the other dialects, this part the accusative which I'm glossing as accusative. So what seems to have what I do not agree with you is that Ra is a typical C, the verb to C is a typical two place predicate subject object. So object, I mean Maria is what's being seen and I guess the most neutral way of considering Maria is to be the object and not the secondary object. Whereas with your analysis I'll be trying to force a secondary object interpretation to Maria which I wouldn't want to. And actually if I substitute Maria by a pronoun I do not get right la but I get righta. So there is no little marking there. Because you're treating this as a prepositional phase. You're treating that as a prepositional phase. Whereas I'm treating it as a noun phase that is accusative marked. It's not a complete verb unless there is a context where something is missing. But this is more tense. The example from Egyptian I would say No, I'm thinking of Baat for example. Batilha for example. Batilha a jawab or whatever. Batilha a qistab. Datlha and if you negate it it will become Mabatil Hish. And that Hish ash or what? So there that can only be treated as a pronominal form and not a PP inside a verb phrase. How can it be? A PP inside a verb form. It wouldn't make sense if you still want to consider it as a PP. Prepositional phase it is. But I mean even Ra'a in Arabic is really a two place predicate. So I wonder how you can force a secondary object reading to the item that's being seen. That's what I I do not agree with you. Cos if this were Shaf Shaf a ha Shaf is also a two place predicate like Ra'a. No? No, I'm talking about dialects, Shaf. And that Ra'a in this right is exactly the same as used in classical. Yeah. The point is in order to say that it will know you in Maltese you must say right little that's how you say it in Maltese. Yeah, but but in dialect. Yeah, Maltese is odd like this. Other dialects are different but Maltese is like this. So differential object marking is only found with object I meant, not secondary object is only present in Maltese. The other dialects do not seem to have it or do not have it. As opposed to the dative, the dative is across. This is accusative though. Tell me. If you go back to the last dialect and look at directional propositions there's two. There's Lila, which is one and then there's Illa. Ewa, ewa, ewa, ewa, ewa, ewa, yes. Now when I actually it was really difficult to find Lila in class in the Quran because usually they are all Illa. Illa, Illa. Thanks for that. Sorry. This is Illa. It's a directional propositions and actually I would say these are even. Can you do the previous slide? Yes. This one B is kind of all action because in that context where it's truly directional you expect Illa usually. Anyhow, that's my point is that there are these two things and they're very similar simultaneously. There's five differences. And then when you get to the dialects now I kind of, when I teach dialects, I say Illa is gone. Doesn't exist. But if you and kind of meaning-wise and functional-wise that's a convenient thing to say. But what you did and we were by no means the first to do it the whole time we were talking about something which is phonologically Illa or Illa and then suddenly we have Illa and you say it's the same. In the universal perfect. You say it's the same thing and like I say you're not at all the first to do this and if you look at Arabic dialects they do kind of synchronically they do seem to function the same way. So like he has his Illa and the negation is Illa or no, not Illa but Illa Anyway this is kind of marginal as a possibility in most dialects. Illa and the negation probably won't involve this in. So it looks like they are part of the same thing but I just wonder are they is it legitimate to really say to say this is just the same thing as there is this historical precedent that's still on the side. Yes. You're right you have a point. What is nice however is that what's beautiful to retain for sure to maintain is the grammaticalisation path because if this were really let's say if it were really the same law used in possessive structures then this is one way in which one could expect grammaticalisation. One way in which one direction which possessive structure could grammaticalise. So kind of it would fit nicely. I kind of suspect that I think one really ought to show this but somehow but I kind of suspect that at some very early point Illa and Illa fused themselves. True. So that would come even perhaps at the peace stage where Illa preposition. Yeah. But yes you have a point for sure. I mean they are more for lots of things screaming. So in classical Arabic it's Illa with a long R but in most dialects that final long round would be shorter anyway. So all you have is Illa versus Illa. We are screaming. And then you typically also find in the east the more east you go and the more gulf you become sar is obligatory. In which case then the vowel drops and you get sarla. So then kind of your left would just a loom. Whereas in Maltese and Syrian I mean at least when it's not with a sar it's clearer. There. Yeah but yes. Thank you.