 Felly, rwy'n meddylch chi'n bwysig. Mae gydag gwasanaeth y flasaf o'r ffordd Llyfrgrwyr. Felly, roeddwn i'n ffordd Llyfrgrwyr yng Nghymru. Mae'r gweithio'r gwaith am gydag i gyda'r ddau gwneud. Mae'r ddau'r ddau honno i gafodd o'r ddau. Mae'r ddau eich gweithio, gyda'r ddau gwneud o'r ddau, o'r ddau'r ddau, oherwydd mae'n gweithio i'w gwneud. I'm sure for the fact that I'm here, I think her teaching has played a big role in my development, I'm sure I speak for others here as well. Aisha has worked on Berber, various varieties of Berber, did her PhD here. And has worked on motion, mood, tense aspect in Berber in the past. Directic Directionals, Association of Motion in Berber and then more broadly in African languages. And today is kind of going back to talk to us about state in particular. So the syntax of normal states in Berber. Thank you. Yes, so the focus of my talk today is an alternation in the morphological marking of nouns that is found in some Berber languages. Berber languages are languages that belong to the Afro-Asiatic film and they are spoken all across North Africa. So from Morocco all the way east, I think, into Egypt and in the south of North Africa in Mali, Mauritania, Burkina Faso and Niger. Sorry, Niger is missing from the list. There is no agreed subclassifications of Berber languages. So it is common to classify these languages by the countries in which they are spoken or the regions in which they are spoken. There are some important lexical and phonological variations across the family. The morphosyntax is more constant but there are some micro variations, of course. So, as I said a few seconds ago, the focus of the talk is the alternations in the morphological marking of nouns. There is an example of what this morphological marking entails in one. So in one A, my stomach, it hurt me. You see the first noun in that sentence, Stomach, contains a vowel, an A vowel, which is glossed as FS. So that's an example of a noun in what we call the free state form. And in one B, my stomach hurt. So it's the same sentence but the word order is different if you look at the noun Stomach. Again, now you see that the A vowel is missing and the gloss says CS and that's for construct state. So the aims today is first to describe the syntax of this alternation. And I'll do that based on data from a few Berber languages. So I'll look at data from Takbalit, which is spoken in Algeria. There are a couple of examples from Tuareg. The variety spoken in Niger and Mali. And then some variety spoken in Morocco. The second aim is to look at a proposal by Koenig that this nominal morphological marking is in fact a case marking, which follows a marked nominative pattern. And the third aim is to explore the link between case marking and information structure. So I'll start with a basic description of the syntax of Berber. I'll start with core clauses. So what I call core clauses are the clauses excluding the discourse function position. So in these core clauses, the verbs can well surface in different stems. There are different stems across Berber languages, but three are found commonly the imperfective, the perfective and the heuristum. These verb stems, they are obligatorily marked for subject agreement is marked depending on the person and number by prefixation, suffixation or both. The unmarked order in these core clauses is VSO. All arguments can be omitted if they are recoverable from the context and if they are cross referenced on a verbal head. So it's either the verb or tense aspect, mood particle, which precedes the verb. And cross reference occurs either by subject agreement. So, yeah, this is either with the subject agreement markers, accusative or dative clitics. Indirect objects, they can also optionally be doubled by dative clitics. And in these core clauses, so post verbally, arguments can occur in different orders. There are a few examples, four examples in two which show everything that I've just explained about the syntax of core clauses. We can look at a couple of these examples. So, for instance, to A, God gave me a fiancée. So that's an example of a VSO order. The verb is first, God is the subject and fiancée is the object. And the verb carries the subject agreement marker, which is a prefix. And it also carries a dative clitic, which occurs after the verb stem. And that dative clitics replaces a sort of nominal indirect object. In 2D, so we can skip 2B and 2C in 2D, she cuts the navel string from the boy. We have an example in which the indirect object is doubled by a dative clitic. So the s just after the verb is a dative clitic, and it doubles the indirect object to the boy. And as you can see from these four examples, s direct object and indirect objects can occur in different orders. Now, alternations to these canonical orders are usually related to information structure. So, topic arguments all occur in the preverbal position. Subjects, direct and indirect objects are obligatorily cross-referenced on the verb by respectively subject markers. Accusative and dative clitics. There are two examples in three. The first one is an example where a subject is the topic. My sister-in-law, she touched me here, so my sister-in-law is moved to the preverbal position and it's cross-referenced by the agreement prefix on the verb to touch. In 3b, we have an example of a direct object which is in a topic. This boy, I carried him in my hands, so the boy again is moved to the preverbal position and it's cross-referenced in the core clause by an accusative clitic, so the t thing after the verb to tech. Focus arguments also occur in the preverbal position. They occur in cleft constructions, so the focused constituent or argument is preceded by a copula, the, and it's phonological variance depending on the verb language and it's followed by what is usually considered to be a complementiser, e, I, or add, again depending on the variety. There's an example in 4, it is tears that I was crying, so the noun for tears is in focus, it's fronted and it occurs between the copula d and the complementiser. In subject cleft, so when subjects are in focus, the verb in the core clause in most varieties must occur in the participial form and it carries default agreement and default agreement in verb are is marked by the third singular marker. So there's an example of this in 5, it is me who bought two books. The pronoun me is in focus here and you can see the verb is in the participial form which is marked by the suffix na and the agreement is a third singular masculine. It's the default agreement. Right, so now I'll discuss the state alternation, the syntax of the state alternation. I'll start with a brief description of how the state alternation is marked morphologically and before that I need to explain a little bit about the morphology of nouns, mainly how gender and number are marked on nouns in the language. So masculine is on marked, the feminine is marked by te which is prefixed and suffixed on to the stem in the singular. So I give two examples in six. The first noun, am rar, old man, is tam rart in the feminine. Isli, groom, becomes tislit in the feminine. The plural is marked either by a change of the initial vowel and a suffix or by a suffix session only. So with some nouns there is no change in the initial vowel. Again there are a couple of examples in seven. So the singular noun am rar becomes im radon in the plural. So the initial vowel has changed. It's gone from an a to an i and there's a suffix an at the end. The second noun, adgar, widower, becomes adgaran in the plural. So the first vowel is actually not changed. In addition to these nouns in most verbal languages they occur in two forms depending on their syntactic function and position. The first form is the free state form and it's from the French Lethalib. And the second form is called in the English literature the construct state form. I'm using that term here but for those of you who are familiar with Arabic and other languages with the construct state, the construct state in verbal is a completely different phenomenon. But I used the term because it's the term which is used in the English literature on the state alternation in verbal. So the free state is the unmarked form of the nouns. It's considered to be the unmarked form because it's the citation form of nouns. And also because the construct state is derived from the free state through changes in a noun's prefix. There are variations across verbal languages as to how these prefixes are derived. But to summarise in the masculine you have either a change of the initial vowel. So usually a becomes u as in aiazid, free state, uiazid construct state form, rooster. Or the initial vowel is deleted and a glide is inserted. So argaz in the free state form becomes worgaz, man. And irgazen in the free state form becomes yirgazen, man. So the initial vowel is deleted and u and word is inserted. For some nouns as well only a glide is inserted. So there is no change in the initial vowel and the example I give is as for day. Sorry I forgot to include the gloss. So it's as for day which becomes was in the construct state. In the feminine only the initial vowel is deleted in the construct state. So tamhart, free state, becomes tamhart where only the a is deleted. The free state because it's the unmarked form of nouns occurs in a wide range of context and it occurs obviously in more context than the construct state. So we find it for citation. As I said before we find it on adjectives or adjectives, adverbs and other sort of context. I give an example in an adjective used in the free state form. But today I would like to focus on only the syntactic context where there is a contrast between the free state and the construct state. So there is a contrast between the two states when we look at preposition and particularly at the complements of prepositions. So in most verbal languages most prepositions have a complement which occurs in the construct state form. So which has to occur in the construct state form. I give a list of prepositions so you have the genitive of sir which is the instrumental. It means with etc etc. There are two examples in 9 and 10. 9, the contraction started at 6 in the evening. The noun evening is the complement of the genitive preposition and it's in the construct state form. 10, she removed the navel string from the boy by cutting it. So literally she cut him the navel string to the boy. The noun boy is in the construct state because it's the complement of the dative preposition e. We also find construct state, well nouns in the construct state when they come after what are called prepositions in the literature. But they're not really prepositions there. I mean they are noun but they have the meaning of prepositions. And these so-called prepositions are always followed by another preposition, the genitive n. And the genitive n can be omitted depending on various reasons. So in 11 he sits on the carpet. The noun carpet is in the construct state because it's the complement of that preposition, nominal preposition on Sufilda. Now there are also some prepositions which do not take complement nouns in the construct state but they take complement nouns in the free state. So all the prepositions in all varieties of Berber, all the prepositions that are borrowed from Arabic like Uqbal before, Mblau without, they always take a complement in the free state. And there are other prepositions specific to dialects that take the free state. There is one which to my knowledge is common to all the Berber languages. It's the preposition meaning until and that preposition always take a complement in the free state. And I give an example in 12. I walked until the camp. You can see that the camp is in the free state. Inside NP we see a relevant-ish contrast between the construct state and the free state. So the head of a noun phrase is always in the free state unless it's in a syntactic context where it needs to get the construct state. So if it's say the complement of a preposition which governs the construct state, it will be in the construct state. But canonically it's in the free state. And nominal arguments such as possessors and complements must occur in the construct state form. And in fact these possessors and complements, they are always, well in all languages, they are preceded by the genitive preposition and of. But that preposition is optional. So in some context it can be omitted. There are two examples in 13 and 14. The example in 14 shows an example of a possessor. So the man's house or the house of the man. The man is in the construct state. 14 we made butter couscous. So butter is the complement of couscous and it's in the construct state. In the, well as far as the subject function is concerned, there is also a relevant contrast between the construct state and the free state. So both in transitive and transitive subjects. So s and a in core clauses. So that's when they follow the verb occur, must occur in the construct state form. There are two examples in 15. The first one, 15a, is an example of an s subject. Now it's over. The boy is born. The s subject is in its canonical position. It follows the verb, it's in the construct state. 15b, the sister of the groom cooked the meal. The sister is a transitive subject. It follows the verb, it's in the construct state. When s and a precede the verb, they cannot be in the construct state, though they have to be in the free state. So in 16a, it is the sister of the groom who cooked the meal. The sister now precedes the verb. I mean it's in a copuloc construction. So it precedes the verb and it's in the free state form. 16b, that song, it was on from the moment we got there. The song is an s argument. It precedes the verb and it's in the free state. It has to be in the free state. Right dislocated subject, whether transitive or intransitive, so whether s or a, have to be in the free state also. And there's an example in 17. The example is from Tarifidbaba, spoken in Morocco. So she soon became pregnant, her daughter-in-law. So the daughter-in-law is an s here, I think. And it has to be in the free state form. In copuloc clauses, we have also a contrast between the construct state and the free state. So copuloc clauses in Berber are marked by the copulod, followed by a nominal predicate. That nominal predicate, sorry, is always in the free state, no matter what. There's an example in 18. Damshi, sorry if I can't speak Berber. It is a cat and the noun cat is in the free state because it is the nominal predicate. An s, so a subject in a copuloc clause can be overtly realised in which case it canonically precedes the d plus predicate complex and it has to occur in the free state as well. So 19, akshisha demich. This child, it is your son. A child is the s argument here. It precedes the predicate and it is in the free state. Direct objects are in the free state regardless of their position in the clause in, well, in most other languages. So there are two examples in 20. 20a, at the time of the lamb's latest harvest, we would go to harvest lamb's lettuce. So at the end of the sentence, there's the noun lamb lettuce. It's the object of the verb harvest. It's in the free state. In 20b, this boy, I carried him in my hands. Here, the direct object is in preverbal position. It's also in the free state. Now I will discuss some micro variations. So most of these micro variations, they concern takbalit, which is spoken in Algeria. So I'll talk about a variation in copuloc clauses. So as I just said, in copuloc clauses an s, so an intransitive subject, can be overtly realised and in most verbal languages it precedes the predicate. But in some varieties such as takbalit, tamazir, tarifit, the s can follow the predicate. In some of these languages, the non-initial s can be in the construct state. That's what we find in takbalit and tarifit. There are two examples in 21. 21a is from takbalit. So that girl is or was mute. The subject, that girl, follows the copula and the predicate. It's in the construct state. 21b is from tarifit. She is an ogres, this one. The pronoun, this one is the subject and it's in the construct state. And in other varieties, the non-initial subject is always in the free state. So in 22 there are two examples from tamazir. This mule is unruly. The noun mule precedes the predicate. 22b is the same thing. This mule is unruly except that the noun follows the predicate there and it's also in the free state. Many varieties of verbs have what are called pseudoverbal constructions. So they are constructions which involve a predicate which has some properties of verbs and which doesn't have other properties of verbs. So these predicates, they usually take some form of agreement like verbs but they cannot be inflected for aspect unlike verbs. And the agreement marker which occurs on these pseudoverb is often an accusative clitic. In these constructions the s argument is always in the free state but in takbalit there are constructions in which the s argument can be either in the construct state or the free state and again it depends on its position. So if the s argument follows the pseudoverb it's in the construct state as in 23 diritwyrgezyna. So that man is bad. The subject wyrgez is in the construct state. And if the subject precedes the predicate it's in the free state. So 24, this grape is not good for eating. The grape is in the free state because it precedes the verb. And finally there are varieties of takbalit and some varieties of tamazirt where a direct object can also be in the construct state. It can be in the construct state if it follows the verb and it is cross-referenced by an accusative clitic on the verb. So a so-called accusative clitic doubling. We can see this in 25. So 25e he adds the meet. The direct object meet is in the free state. That's the expected marking. In 25b he adds that meet however the noun meet is in the construct state. It's in the construct state because it follows the verb and it's doubled by the clitict which occurs on the verb. The direct object which is doubled by a clitic and in the construct state form may occur inside the core clause. That's what we have in 25b or it can occur after a prosodic boundary. So it would be outside of the core clause. That's what we have in 26. She found it was the mountain cat who inhabited the house. Who inhabited it the house. So the noun, the direct object, sorry, house is in the construct state because it's doubled by an accusative clitic but it occurs after the clause boundary. So in 27 I have a little table summarising the context in which we find the construct state and the context in which we find the free state. I won't go through that again. But I think basically what the table shows is that there are just two main contexts where we find the construct state is when the noun is a dependent of a preposition or when the noun is a dependent of a verb. So this is basically what we have with case marking. Case marks, nouns which are dependence on different heads, usually a verb and a preposition. So the state alternation in verb looks like a case alternation or some kind of case marking. However, there are not a lot of case analysis of the state alternation amongst the verb literature. I'll skip some detail about the literature a little bit. The majority of analysis on the state alternation they focus on word order and the sort of syntactic or structural configuration in which the noun occurs. So for them the construct state, a noun is marked for construct state if it occurs next to a particular head. And there are other kinds of analysis which analyze the construct state as marking some particular functional relations between the noun and a preceding head. There is a proposal not in the verb literature made by Koenig and Ikenwald and they propose that the state alternation in verb is actually a case alternation and that verb languages have what they call a marked nominative case pattern. And that's what I would like to discuss next because I think it works for verb. It's a good analysis of the state alternation. So according to their argument or analysis verb languages follow a marked nominative case system. So marked nominative languages are typologically very rare. They occur mostly in Africa and they are a case system that share characteristics of both nominative systems and ergative systems. So like nominative systems they treat in transitive subject S and transitive subjects A in the same way and they contrast them with direct objects, so O. And like ergative system the form of the transitive subject is the one which is marked. And I give two examples in 29 of a language with marked nominative. The language is tenet. It's a Nilo-Saharan language. So in 29A, sorry, squirrel went there. We have an S which is marked with the nominative marker E. 29B, lower-speared YOMA. We have a transitive subject lower which is marked with a case E. And an object YOMA which is on marked. So Konig 2008 reviews in detail seven languages which she argues follow a marked nominative case system. Most of these languages, in most of these languages, sorry, the nominative and the accusative are part of more complex case paradigms. So the markers that they analyzed as marked nominative and accusative are on controversially case, not like in Berber. And so in 30, I don't have the time to go through the entire description of marked nominative. So in 30 I have summarised the context in which we find both the accusative and the nominative in marked nominative languages and I contrast them or compare them to the context in which we find the free state and the construct state in Berber languages. So the idea according to this marked nominative analysis is that the accusative corresponds to the free state or the free state is a marker of accusative and the construct state is a marker of nominative. And as you can see from the table in most context, I mean the free state behaves exactly like the accusative and the construct state behaves exactly like the nominative. And there are some contexts which are particularly interesting, I think. So the S and A in postverbal position are never marked with the accusative in marked nominative languages. They are never marked with the free state in Berber languages. They are always marked with the nominative in marked nominative languages and they are always marked with the construct state in Berber languages. The S and A, when they are in preverbal position, on the other hand, are always marked with the accusative in marked nominative languages and in Berber they are always marked by the free state. ac ydych chi'n eich bwysig arlawni'r ffordd arlawni yn y ddau cyd-nw. Ac yn ddau cyd-nw y ddau cyd-nw yn burba'u lluniau. Felly mae'r proposol yn ymgag yw i'w gynyddoedd yr oed ar burba'u cyd-nw. I think it's because there are some arguments, well, there are some issues that are not addressed and these issues are usually the main arguments that people in the Berber literature give to analyse the states as states rather than case markers. One of these arguments is the word order and precedence, so the construct state is only found when it follows the head that it depends on. Another thing is the mismatch between the state and the grammatical functions of S and A. So the argument against coning analysis is that if the construct state is a nominative case then in transitive and transitive subjects should always be in the construct state regardless of their position. And the third argument is the mismatch between the construct state and the grammatical functions of object and clitic doubling in some varieties. And what Berber scholars usually say is that if the construct state is a nominative case, sorry, it should not mark direct objects and they also focus on the fact that the construct state on the object goes with clitic doubling. So for them the clitic doubling is the thing which triggers the construct state. Right, so in the last few minutes of the talk I would like to provide some kind of explanation, I want to keep coning's proposal and I'd like to propose some kind of explanation for the differences in the case marking or the state. Marking, so my hypothesis is that the construct state and the free state are case markers so the state alternation is a case alternation. The construct state is the nominative, the mark nominative and the free state is the accusative which is the unmarked case as proposed by coning. And the discourse functions that are associated with S and A or the discourse functions with which S and A are associated in Berber and the O in Tachvalid trigger these changes in case or in state. So I first look at the word order or precedence argument, so the argument which says that the construct state, well, the construct state has to occur because it follows ahead. Well, in fact in Berber when we have changes in word order outside of the core clause these word order changes they are related to information structure all the time and they occur to mark the discourse functions of topic and focus. And these two functions of topic and focus are, well, highly associated with the preverbal positions. Even in copular clauses the initial position for an intransitive subject for NS has been shown to be a topic position. Where am I? So in fact when a noun is marked for topic or focus it has to precede the verb and if a noun precedes the verb then it's either a topic or a focus. So we can restate the precedence in information structure, sorry, as in 31. So nouns that are in discourse function related positions are in the unmarked case which is the free state aka the accusative. So a noun is not in the free state because it doesn't precede ahead, it's in the free state because it's in a discourse function related position. 31 is actually supported by the fact that both transitive and intransitive subjects, so both S and A, are in the free state form when they are right dislocated. And we saw an example of that in 17. But I won't go back to it. The second argument or the second argument against Koenig's proposal mismatches between states and grammatical functions of S and A. Well, the transitive and intransitive subjects are always marked, so they are always in the construct state when they occur in their canonical position, so when they occur after the verb. But that's because it's their canonical position and the only context where we don't find a construct state on an S or an A is when the S and A is in preverbal position when it is marked for either topic and focus. So we could say something like 32. S and A in transitive and transitive subjects are marked with the nominative case aka the construct state, but they are unmarked when they are topic and focus. So when they are in discourse function related positions. Now the last thing I would like to discuss is the nominative or the construct state that we find on direct objects in a clitic doubling context. I think but I, okay, so this is a hypothesis and I don't have clitic doubling in the varieties that I work on, so I don't have the data, so I based myself on the data that I found in the literature. I think the objects marked by construct state in clitic doubling context can also be related to information structure. First the O in these constructions so the object in those constructions must be definite, but when it's not doubled and not in the construct state it doesn't have to be definite so it can be either interpreted as definite or indefinite. So verbal languages don't have definite or indefinite determiners, it depends on the context whether a noun is interpreted as indefinite or definite. In 33 he had a suite or the suite the direct object is not doubled by a clitic. In 33b on the other hand the direct object suite is doubled by a clitic and only the definite interpretation is available, so 33b can only meet he at the suite not he at a suite. Secondly a direct object in when it is in the construct state and doubled by an accusative clitic must have a reference which is identifiable. And this restriction does not apply with a regular direct object so 34a he at I don't know what we have well it's not really an object well it's not an object well it's doubled anyway. We have something which is not an object which is a complement clause I think. And well it can be the complement of the verb to eat in 34a it cannot be the complement of the verb to eat in 34b because the verb is doubled by the complement clause is doubled by an accusative clitic. Although yeah sorry I'm not sure about that example anymore it's a bad example sorry. Well anyway yeah so the fact that the object has to be definite or identifiable does not show alone that the object is topical but I mean usually if something is topical then it has to be definite. Or well often often it will be definite and identifiable. But there are other facts supposing the hypothesis that a clitic with a direct object sorry which is in the construct state form and doubled by a clitic is topical. First in all verbal languages and we've seen some examples at the beginning of the talk direct objects which are in topical position they must be cross referenced or double by accusative clitics in the core clause. So there is a link between clitic cross referencing or clitic doubling and topicality in verbal. Second while accusative clitic doubling is rare it only occurs in Takbalit and some varieties of Tamazir there's another clitic doubling that occurs across most verbal languages it's dative clitic doubling there's an example of that in 35. Now dative clitic doubling in verbal has been shown to be linked to information structure it's been shown by Kosman and Swag. And finally in Takbalit when we have a right dislocated object which is obligatorily doubled by an accusative clitic it's a topic so that yeah it's a topic it's a topic which is reactivated. So we can explain the link between the construct state and the clitic doubling and link it to information structure and this is summarized in 36. So o in verbal is unmarked in Takbalit o can be simultaneously doubled by a clitic and marked for nominative case if it is topical and follows the verb. Now as I said I'm missing a lot of data and this is only a hypothesis. And nobody really has explained the state alternation or the case alternation in verbal in terms of information structure but some people have linked the marking of nouns to information structure. So like Fiwi 2014 she focuses on topic constructions and she considers that the free state participates in determining a topic. And Metushi and Flesh they also link the difference in case and state marking to discourse functions but they do that from a syntactic point of view. So their hypothesis is that at some stage topic and focus arguments were outside of the clause and so they could not receive canonical or normal case marking. I at this stage cannot explain why the information structure would affect the case marking of S, A and O in such a way. And why a topical O would have to be marked simultaneously by the construct state and a clitic doubling. However the alternations in the case of subjects and direct objects depending on either semantic or pragmatic factors do exist in other languages and they are reported in the literature. For instance Derrimpol and Nikoleva they describe a similar ish phenomenon for direct objects in two symmetric languages. And in these two symmetric languages a direct object is normally unmarked but it can be marked for case when it is topical. Derrimpol and Nikoleva also show that optional object agreement can be linked to topicality. So assuming that clitic doubling is a kind of agreement then we can say that topical objects in takbalit have to be marked or can be marked optionally by both agreement and case. So the conclusion. The syntactic distribution of the nominal state alternation in Berber is similar to the distribution of nominative and accusative cases in marked nominative languages as suggested by Koenig. At the closer level the state alternation marks the grammatical functions S, A and O, the unmarked form, the free state or the accusative marks all arguments always in the canonical clause except in takbalit, the marked form marks subject and transitive and intransitive subjects. It seems that the discourse functions of these grammatical functions trigger changes in the canonical case marking or the canonical state marking. S and A are unmarked if they are in topical focus positions and in takbalit a direct object can be optionally marked if topical. Now there are two domains that I haven't looked at today but I wouldn't have had the time anyway. The noun phrase and the prepositional phrase, I don't know what happens, I don't know what happens in the noun phrase and the prepositional phrase might be a bit more, might be different. I think I'm done. So here we have loads of questions, comments, this is kind of a big experiment. Thank you very much Aisha, I very much enjoyed that, where I was stimulating. Do you have any thoughts about the diacramy of this because maybe I just kind of dreamt this but I have this vague notion that some people have tried to suggest that at least some of the free state morphines might be derived from demonstratives All definite articles, I think that's the idea. No, it's just because if that's true it does seem to me to my mind explain quite a lot because you expect subjects to be definite and you don't expect objects to be definite Cross linguistically you commonly get this phenomenon of differential object marking, if an object out of character is definite then you have to do some extra signalling of it. Then also possessors tend to be definite so I like your analysis and I'm not in any way disputing it but I guess I'm saying I always like to look at things diacronically and it would make sense to me if it had, the free state was originally some kind of definiteness marking because that could explain a lot of the distribution we now see would be consistent with the what you would expect when it was a definiteness marker in the past. Maybe I don't understand the definite so if the definite would explain why a subject is marked? It's the kind of logic that would have been flipped but originally what would have happened is that subjects tend to be kind of old news and therefore definite so it just would have happened that very often subjects would have been marked with this definiteness marker and then as it lost its kind of special definiteness meaning then the kind of motivation flips and it becomes a kind of signal of subject. But I need to check because I think it's the free state vowel which people argue derives from a definite marker. I think it's the R and the R is, I mean the free state is not the canonical subject marker but the word, you still have a good point because the word as well. No, no, no, it's not a dream because I know I've seen it so you haven't dreamed it. Yeah, follow up slightly in a different direction but I think pretty much along the same lines and I also wanted to thank you. Really fascinating data, I'm not sure I've got all of it into my non-verbal head. So going back to quite one on page A where you systematise what a marked nominative case system looks like and then it combines properties for both nominative and additive systems. That is really interesting because it immediately reminded me of John Ludwyr's preferred argument structure which is not outdated so I don't think many people know. No, I'm not familiar with that. So that is, he's written several, one language paper and several other papers on the discourse basis of, on the cognitive basis of discourse which means that even in nominative, accusative languages you get a distribution of full MPs with this chronominal MPs that reflect cognitive systems. So that you have S and P as full MPs and A chronominal and that is of course related to information structure as well. So in discourse you wouldn't get the man hit the boy, you would have something like the man came, he hit the boy. Yeah, so this is about the distribution of the numinal MPs. And so it looks a little bit like verbal languages conflate that so nominative accusative languages differentiate. They have one in their case system and one in the information structure and it looks a little bit like verbal languages might conflate this by paying attention to information structure criteria and for medicalising them in their case system at their first sight. Okay, and that's Du Bois. Du Bois. What's the first name? John. John, UCSD. Thank you. Other questions going to that side? I was just wondering, when you have this mark or object in 25G, quite confusing, the way that the alignment works. So I was wondering if the object is marked and the subject would also be marked in the same way. Am I saying it the wrong way around? Ah yes, if you have a subject and an object, yes, yes, yes. The subject would be in the construct state and the direct object in the construct state. Do you know marking in mathematical relation? That's what you're saying. No, I'm saying it's marking grammatical functions. Okay. Except in this context. Yes, yes, yes. We agree. Out of curiosity, why are the researchers working on verbal stuff reluctant to take up the case analysis? Is it just the case of an established regional school? No, no. Do you think it's very common for us in the big languages? First of all, there's the word order thing. It's also, there's some sort of influential scholars on verbal, basé, gallant, shaker, and they don't want a case. I mean they don't want to analyse that as a case. They want to analyse that as a structural dependency. Okay. Yeah. Plus the prepositions as well. So these prepositions that take the construct state, not the nominal prepositions, the other prepositions. In many languages, particularly in Takbali, they tend to attach or combine with the noun. So they form just one word. So there are a few analyses showing that. And also the construct state is formed by deleting a vowel or reducing a vowel. So the argument is it reduces phonologically. So it's phonologically deficient, hence syntactically deficient. Hence it needs to be close to something else. It's a combination of things. Ignorant as to your work, but where are your data from? So the Takbali data which is not references mine. I collected it when I did my fieldwork a few years ago. And the rest of the data, so all the data which is references from the literature. So either grammars, descriptions of information structures. So a kind of a follow-on from Chris's question, but in the other direction. So I know some people working on the state in various varieties have discussed it being eroded. The state alternation being lost. Perhaps through concept learners. And I was wondering if in that context you have anything to say about which of the alternations are lost, if we know. The construct state. So everything then collapses to the free state. And there is a particular context in which that happens in more than others. Or whether it seems to be rather complicated. So some are more robust than others. But I don't know a lot. I know that there are traces. So all the Libyan varieties, Egyptian varieties have lost it. There are traces in the prepositional phrases. So some prepositions still give the construct states to their complements. And there's also, yeah, also in the south of Algeria, Zab has lost the alternation. But yeah, in the prepositional phrases you can find some traces. So the traces are, as far as I know, they're always in the prepositional phrases. But I haven't really looked for it yet. Mark, the nominative systems across the industry are very rare. And what I find striking is the aerial distribution. Because they are tested in all verbal languages. And then in East Africa in a number of languages of different genetic relations. Not in Arabic, interesting. So you wonder what props up such a system. I don't know. I don't know. The Nilo-Saharan languages, I mean verbal languages, have many similarities with Nilo-Saharan languages. But the Nilo-Tic languages, often there are many, many similarities. But I don't know what it means. I'm going there. You said one of the arguments against the case now. Is that the barbarous, usually say the contract state is deficient? But if you look at the phonological processes that are involved and the morphological processes, it's always like an augmentation or like building upon the free state. It doesn't ever seem, from what I've seen in the literature and the phonology at least, that the contract state is like diminished. A lot of the analysis there in the government phonology framework, CV, CV, CV. So the argument is that the vowel in the free state, so the a thing is long. And so it links to two CV, one CV, CV, two, well CV, CV. And the contract state vowel, they disappear. So we lose one CV. So it's deficient. I mean, to me that smells a bit funny. I was convinced. I don't know, I read them, but it's funny to have them have phonological arguments for syntactic and morphological. Because in the prepositional domain it kind of works. For prepositions it kind of works. Not anywhere else. Questions or comments? I think we can thank Aishu again for those comments. Thank you.