 Owen did his BA honours at A&U in Canberra and worked on Talaki from the South-Eastern Sulawesi in Indonesia and he's just submitted his PhD dissertation supervised by Mark Donnelly Thank you. So in Australia we don't have a, they don't have, I should say they, a system of biver like in the UK, so you don't actually have to defend your visitation, you simply send off and the examiners do with it what they will, so Owen's in the precarious situation in a way to hear what the examiners think, but in the meantime he has a endeavour postdoctoral grant to do some field work, more field work in Timor and he'll also be doing some work in the Netherlands on language contact between Austronesia and Austronesia with Marianne Klamer. He's published on several topics, historical reconstruction and Engano, which is a really interesting language, working on the barrier islands just off Sumatra in Indonesia, so the lucky to have him here today, those of us who've come shall hear about Methodesis Timor. Yeah, so I don't know what the title was when I sent it out, but this ended up being the title of Methodesis, this rather grandiose sort of title, so if you're feeling inspired or terrified by the end of the seminar, hopefully that will be online somewhere in a couple of, I don't know, hopefully not too long, a month or so. Yeah, so I'm talking about Methodesis in Amarasi, which is a language of Timor. Oops, I'll just use this keyboard. So first of all, so this is a structure of my thought, introduction, language background, what is Methodesis? Then I'll describe the structure of Methodesis with more things than just Methodesis occur than the functions and we can identify sort of three distinct uses of Methodesis in Amarasi. There's one instance where it's completely phonologically conditioned and completely predictable, which is as well, sort of goes from the phonology, a bit higher to syntax all the way to the discourse. And then I've got some conclusions, Methodesis within linguistics, and then some more, that sort of sort of general linguistics conclusion, then some more Timor specific conclusions, Methodesis and identity, and linking it to sort of the whole culture of the area. I end up with this rather sort of, as I was sort of getting to the end of my thesis, my conclusion was becoming sort of esoteric and I was feeling as though I was making it up, but when I presented a version of this at ANU everyone thought, okay, so anyway, let's get started with the introduction. So Methodesis is when two segments of a word change positions. This is some data from Ratouman in the Pacific. It's sort of one of the most famous cases of Synchronic Metathesis. So nearly every word has two forms and depending on the vowels and consonants, these forms are formed by a Metathesis. There's some other complications as well, so like if you have high vowels, you can get umlaut, so a word like, what's the example like? Like furri becomes fur, and this has a morphological function in Ratouman. It's linked with definiteness and number. I can't remember the exact details, but one of these is like definite plural and the other is indefinite plural. Now, most sort of descriptions of Metathesis, they're focused on the phonology because it's strange and it presents challenges for models of phonology, and there's been lots of good phonological work done, but one of the central points of my thesis is that in Amorasi at least, we need more than phonology to understand Metathesis. So where is Amorasi? It's on the island of Timor. It's part of this Wubmetoclust, also known as Darwin. Timorese, I think Bob Blust calls it a Tony, and Amorasi is at the sort of southeastern end. It's a sort of, it's like a language dialect cluster, sort of like, you know, in German or romance or something, although maybe not as diverse. So here's Amorasi, and I worked in a village called Neckmesse, which is, I think, about there. It's got about two main dialects, but I won't go into that and just presenting on, from the one dialect. So first of all, the structure of Metathesis, what happens? Before we can sort of talk about what happens, we need to know some basic phonological facts about Amorasi. Thirteen consonants, five vowels, these voiced obstruents, they have a limited distribution. They occur in loan words and also in sort of specific morphophonological environments, which I'll get to later in the presentation. And we have very constrained word shapes. So, you know, most words are two syllables, I think, in my current database, about 67% of words, in fact, are two syllables. So here's Fatu Rok, if you've done anything on Austronesian, you'll recognize this. And all vowel-initial words begin with an obligatory glottal stop. So, Umea House. We also have word final consonants, Muith, Animal, and about 43% of words in my database end with a consonant. So that's pretty common. We can have sequences of two vowels, like Kwan Village, that's also quite common. We can also specifically have this sort of diphthong, which I've written here, only in stressed syllables, so we get Nautus. Otherwise, sequences of two vowels are always formed two syllables. So specifically in stressed syllables, we can get this diphthong, and it can be of a number of vowel qualities, but not just any old vowels. We have consonant clusters as well, Khaum, meaning Ratten. This is fun, coming from Southeast Hiloasea, where everything was consonant vowel, consonant vowel. I enjoyed that. And those are also quite common, about a quarter of words, begin with a consonant cluster. And we can have words of three syllables, four syllables. Four syllables is about the maximum, and nearly all four syllable words are probably hysteric compounds. So based on this data, we can say that the structure of an Amorasi word is something like this. We've got a foot, then we've got, which is all optional, then we've got two vowels, which are obligatory, the penultimate one is stressed, and we've got an obligatory initial consonant. If there's not one specified, we get the glottal stop, like in the word for house. And then we've got some and then we've got some optional consonants. Part of my analysis is to propose that actually this isn't the structure of an Amorasi word, but all feet are obligatorily consonant vowel, consonant vowel, consonant. And when we don't have consonant specified, we have empty, empty C slots. So there's sort of null consonants. So now I can talk about the form of metathesis. I'm going to refer to the two forms, the unmetathesized form and the metathesized form as the U form and the M form, because more than just metathesis takes place. With our words that end in vowel, consonant, vowel, we get consonant vowel metathesis. And this is regular, no matter what the vowels are. I forgot two identical vowels, we also get metathesis. Words which end in a consonant have metathesis and their final consonant is deleted. Now if you'll remember from our word structures, well you won't because I didn't say it. We have no final consonant clusters. So this we can analyze as being metathesis. Then we get a word final consonant cluster, which isn't allowed. So you resolve that by deleting the final consonant. We have some vowel assimilation, so mid vowels get raised to high after the high vowel, after metathesis. Again, in Amorasi, we have no sequences of a high vowel followed by a mid vowel. So this also can be analyzed as a straightforward phonological repair. Some other varieties of Wabmeto do allow Wem but not Amorasi. We have assimilation of the vowel R and this only occurs after metathesis. So Nuka, we get metathesis and that R assimilates. We have sequences of vowel plus R in U forms, but not in M forms. So this is a derived environment effect. You have one rule which operates and that triggers another rule. Our words which have a diphthong, they form the M form by deleting the final vowel. Again, we can analyze this as straightforward phonological repair. You get metathesis, then you have three vowels which isn't allowed, so you delete the third vowel. If we've got a diphthong and a final consonant, then we delete the final consonant and vowel. Again, this is straightforward. The sort of most disturbing form is our forms that end in a vowel sequence and a consonant. They form the M form by just deleting the final consonant. There's no surface reason why this should be the case. Kaut is a perfectly good word, as you can see by fault, and there's no apparent reason why that should happen. For the sake of completeness, words that end in a vowel sequence don't distinguish between the U form and the M form. Copying a vowel, what do you mean? If you're switching the position of the vowel and the consonant, there's no vowel to switch. Yeah, but still, from a surface point of view, why does that final consonant get deleted? You've got a morphological paradigm to fail. These ones can't do metathesis, so they need to do something. Now I analyze this as underlyingly metathesis of an empty consonant. We can analyze all these forms by proposing that actually, underlyingly, they're all consonant, vowel, consonant, vowel, and then these get metathesized and any final consonant clusters get deleted. How does this work? Here are our simplest examples. We've got Muit, Kaut, Fatu and I, and I'm going to fill these empty consonants with this null sign just to make it explicit that they behave identically to other consonants. Then we have metathesis taking place. Now we've got a final consonant cluster, which isn't allowed, so we delete that final consonant and we've got our M forms. So that's the sort of, oh, we've got a few more complications. We've got this assimilation, which only happens in M forms, so we get metathesis and then the features of the stressed vowel spread. We can sort of explain the reason why this happens by the fact that R is a sort of featureless vowel in Amorasi. It's also an epithetic vowel, so we can say that it's minus high, minus front, minus everything, and so then after metathesis it's just sort of got no features to sort of stand against the spread of the stressed vowel. In other varieties of Wabmeto, all vowels assimilate after metathesis. In Amorasi, it's just R, and then we delete our final consonant, but that is no surface effect. In words with the diphthong, the same thing happens. We have consonant vowel to vowel consonant metathesis. Now we've got a sequence of three vowels, which isn't allowed, so that third vowel gets deleted, the others shuffle across, and then we delete the final consonant and we've got our M form. So the big generalization that allows us to sort of describe all the different M forms is this obligatory consonant, vowel, consonant, vowel, consonant for every word has that structure, and when you don't have a surface consonant, it's just empty, but it behaves in the same way as though there were a consonant there. So where does metathesis occur? What are its functions? There's one instance of phonological metathesis, which is what I'll start off with. So the number of vowel initial in clinics in Amorasi, and whenever these attach to a clinic host, that host obligatorily occurs in the M form. Now these in clinics form a phonological word with the host. Stress sort of remains on the host, but syntactically they form their own independent syntactic head. So for instance this determiner is the head of a determiner phrase and the noun to which it attaches to is the head of a noun phrase. So there's this mismatch between the syntactic and phonological structures, which explains some of the processes. Now some of the forms are a bit different before vowel initial in clinics. This form is the same, we have a simulation of R, but word final consonant survive because they're no longer after metathesis, that consonant cluster isn't word final, so no problem. Similarly for our words of the diphthong in a final consonant, that consonant survives. These words undergo no difference. Vowel final words, we have a consonant appearing, we have consonant insertion. So a word for Paul, we get knee jet, and this is the environment in which the jet and the gua occur. Now which consonant is inserted is phonologically conditioned. After front vowels we get jet, after back vowels we get gua. So this is you know you know this is phonologically natural, after back rounded vowels you get a back rounded consonant, after palatal or front vowels you get a palatal consonant. When the the vowels of a word are of a different of different qualities then we get consonant insertion and the vowel which conditions insertion of the consonant then assimilates to the quality of the previous vowel okay. So I fire a jet and this is regular so in depending no matter what the vowels are, oe, water, oje, kiu, kigwe, ao, agwe. And so the contrast between lime and fire before a vowel initial and clinic is maintained but it's moved off the vowel to the consonant and there's a bit of collapse here because a word which ended in a u would be pronounced identically okay. When words end in a vowel and a consonant we get the final vowel conditioning insertion of a consonant then we get metathesis and then the vowel which conditioned insertion of the consonant assimilating okay. So it's a fafi, fafje, ome, omeje, so on and so forth. Yeah so this is kind of disturbing but surprisingly straightforward to analyze once you know how it works. I mean I spent you know three years figuring this stuff out. So we've got three processes before vowel initial and critics. We've got consonant insertion, we've got metathesis, we've got vowel assimilation and I propose that all of these are analyzable as due to automatic phonological as automatic phonological processes of the specific environment which they occur before a vowel initial and clinic. So first of all consonant insertion. If we take a word like ome, house, this is the sort of structure, we attach our and clinic. Now consonant insertion happens because morphemes need an onset consonant. This is a common constraint cross linguistically you want to begin with a consonant. So we've got these two morphemes neither of them begin with a consonant. The word initially we get a glottal stop so the default word initial consonant is the glottal stop. Word medially we get our consonant by vowel features spreading. So the coronal features, the front features of this vowel spread, we get a jet and now both our morphemes have an initial consonant. I should say also that I'm analyzing these consonants as ambicellabic so it's both the coder of the previous syllable and the onset of the following syllable. Metathesis occurs because we want to separate the two morphemes. It's a crisp edge constraint. At the moment this consonant is shared between both the clinic host and the inclinic. It's a fuzzy border that's not so preferred and so the way we get rid of that fuzzy border is by metathesis. So metathesis occurs now we've got two consonants in that two final consonants in that syllable which is not a good syllable structure so that final one delinks and we've got a crisp edge. It's clear where the clinic host ends and where the clinic begins. So metathesis sort of happens to push the final consonant of the clinic host into the inclinic and vowel assimilation I analyze is occurring because of metathesis. So metathesis has happened and now the features that are shared between the vowel and this consonant that we've got lines crossing. They're shared across the intervening consonant and lines crossing is sort of the cardinal sin of linguistics and syntax, phonology everything. So the way we deal with this is by delinking that that final vowel now we have an empty vowel slot into which that stressed vowel then spreads and we're done. So I'll illustrate with another example this one's got no surface consonants but every single empty C slot I play I posit plays a role in deriving the M form before in critics. We attach our inclinic we need morpheme initial consonants we get a glottal stop word initially vowel features spread word medially we want a crisp edge between the the clinic host and the inclinic so metathesis takes place we've got our crisp edge and now these features are spread across the intervening consonant in this case the consonant is specified as sort of minus everything minus I've put minus C for minus consonantal there we're not allowed lines crossing so that vowel gets deleted and then their previous vowel spreads and we've got our M form before vowel initial in critics and this works for all the different word forms so there they are on the right we attach our inclinic we can fill our empty C slots with the null sign to make it explicit that these behave the same as full consonants consonant assertion occurs metathesis occurs to give us a crisp edge uh vowel assimilation occurs because the blue consonants sort of the sorry the green segments are sharing features across the intervening blue ones so those vowels get deleted the other one spread we're not allowed three vowels so in the form for beetle the third vowel gets deleted and then we've got our final R assimilation which occurs in M forms and we're done so phonological metathesis we've got a number of different processes associated with metathesis and all these are analyzable as due to sort of automatic uh due to the phonological phonological environment in which they occur uh consonant assertion morphine's in an onset consonant consonant metathesis due to crisp edge you want to keep those two morphine separate that's because they form different syntactic heads okay so then the syntactic structure is sort of having an effect on the phonological structure then vowel assimilation because of metathesis so the metathesis element of of this is syntactically driven it's because you're sort of determiner is ahead of a determiner phrase you're now ahead of a noun phrase but it's phonologically conditioned before consonant initial in clinics you don't have any of these processes happening we also have a morphological process of metathesis which is the only signal of syntactic structure uh here's two nearly identical sentences in Amarasi got fatu koal and fault koal which differ only in the in the form of the first uh first noun well they have different structures the form with the u form we've got two noun phrases the form with the m form we've got a single noun phrase uh I should say that there's no evidence for a distinction between adjectives and nouns in Amarasi sort of analyze them all as nominals so in the syntax uh metathesis is a kind of construct form uh this is common in Arabic uh dialects when you've got an attributed modifier you mark it more for morphologically in some way this is what metathesis is doing in Amarasi um yeah here's the extended nominal phrase that I propose for Amarasi and this is where metathesis happens within the noun phrase if you've got an attributed modifier here then the head will be in the m form modifiers numbers determinants quantifiers don't trigger metathesis similarly possessors don't un don't experience undergoing metathesis uh because of the possessive it's only within within the noun phrase and bar also for verbs the same thing this is in a serial verb constructions analyze serial verb constructions as being a kind of um being of the same structure but in the verb phrase uh but let's talk about the noun phrase first uh here's a sort of real-life textural examples example uh this is from a text where someone is describing how you can make a magical sign to protect your garden from theft so he starts off in pake soko her tani tie a rope what kind of rope that's a rope made from the geowang palm so our compound there well is it a compound is it a i'll be our main agnostic it has this structure we've got two nouns within a single noun phrase so the head noun occurs in the m form uh and we can't have the u form here tani tuni is nonsensical that would mean something like ropes are geowang palms um it doesn't make sense also multiple modifiers so every modifier sorry sorry every noun except the final one will occur in the m form so here we've got another example from a text i'll talk about uh uh abit inhabitant field work it's sort of a map inhabitant field work with this structure sort of happily branching up and all the non-final nominals are in the m form also if we have an alternate structure where we've got two nouns here modifying the head noun the same the same thing happens so so so when you think about this you're a person who's an evil doer so it's this is uh a how does it work it's a um evil doer person something like that so anyway this this is how i analyze the structure and this noun is in the m form because of this following one this one's in the m form because of the following in clinic and then this complex noun phrase uh or compound modifies that noun and so they're all in the m form yeah so in the syntax uh m forms are a kind of construct form marking a following attributive modifier of the same word class as the head this contrasts with possession so here's a typical possessive phrase in amaralcy a thorny mthani person's rope uh here's the structure we've got our possessor we've got this optional possessive pronoun and then we've got the possessum and you can't have any of these elements metathasized that's ungrammatical um and this possessive pronoun is optional so that can be deleted a thorny mthani is also uh completely grammatical and so then this contrasts with bainthony the only difference between the possessive phrase and the modified and the attributive phrase is the metathesis of the initial noun they have the same stress same intonation only the metathesis signal signaling the difference in syntactic structure another example in which it's only the metathesis which could be doing it is ordinal and cardinal numbers so card uh ordinal numbers are a kind of noun cardinal numbers are not so ordinal numbers occur within the noun phrase so we get metathesis on the head noun uh cardinal numbers occur outside the noun phrase and so we don't have metathesis on the noun again these have the same same stress same intonation uh yeah so this is this is where cardinal numbers are and ordinal numbers are down here uh also in serial verb constructions we have one verb modifying another so the first verb occurs in the m form here's another example this came from the end of a text where uh my informant had been recording some stories and then the guy who was telling the stories and my informant they both needed to use the bathroom so they left but he'd left the recorder on the table and so the other people were sort of sitting around and they're trying to turn off the voice recorder uh and this one guy's trying to turn it off but he doesn't know what he's doing so this other guy his friend sort of reprimands him and says you're just pressing the buttons of that randomly and he goes on to say just like you do with all those mobile phones that you don't understand so here the serial verb construction is uh it's signaled by the the first verb occurring in the m form so we've got the same sort of structure as a noun phrase just four verbs yeah yeah so within the syntax um metathesis marks kind of construct form now you never get an m form in the syntax without some following u form so when you get an m form it's sort of somewhere not at sort of towards the beginning of the noun phrase you might have a string of m forms and then this u form will close them close it up and say well that's that that phrase is done so we can analyze the m forms and u forms are sort of forming a complete whole a u form by itself is sufficient but an m form by itself is not it's sort of incomplete it needs a u form to be finished discourse metathesis now discourse metathesis so syntactic metathesis that affects both nouns and verbs discourse metathesis affects a wide variety of word classes but most of my examples are of verbs because that's where it's most prominent okay and before yeah so four verbs the m form is the default form okay so if you get a word list all the verbs will occur in the metathesis form so that's the citation form this is different to nouns where they occur in the u form so for instance this this verb for to elbow someone that will be cited as and seal elbows whereas the noun will be cited as silk okay so we've got a difference in which morphological form is doing is marking which semantics so for nominals the u form is the semantically unmarked form for verbs the m form which is the morphologically marked form is a semantically unmarked form and then the m form for nominals signals the marked semantics construct form and the u form for verbs signals some kind of special discourse semantics so what is this sort of special semantics it's unresolved so when you get a verbal u form it signals there's more coming um this is a sort of extract from my thesis of verbal u forms are used by speakers to signal that they do not consider the event or situation encoded by the verb resolved so verbal u forms leave the audience in a state of mild suspense with the speaker signaling there's more coming uh so if you use a u form you're saying you know hold on to your seats there's more coming and there's a number of constructions in which these typically occur one is in coordination so you have one event in the u form a coordinator and then a following event which is dependent on that one so you've got a sequence of events so this is from a text about the creation of the world there was a sort of mythical snake who created the world and he makes and goes out onto the dry land so here the making is only one part of a more complex series of events and you can't have the m form here i've tested this with speakers and it doesn't make sense it's nonsensical um so we've got one event in the u form with an optional coordinator and then a second event and both these events then make up a whole sort of complex event but it's not simply the case that it's some sort of morphological conditioning that you know these uh coordinators require an unmetathasized form because we can get the the m forms before them so here we've got so the old woman spun thread doing it like this now here the event on either side of the coordinator refers to the same event the second verb is anaphoric reference to the first yeah so another construction in which these typically occur typically occurs in tail head linkage so here's another example from a text when it's part way through a text and me and roni my main informant had been uh going around to various places and we finished up at one place and we're going somewhere else so he says and we ran or i said now there on the way while we were going now oh and in a pure and more oh and tanker chief fell it fell and was lost so this is a sort of critical point in the story when i lose my handkerchief so here we've got my handkerchief fall it fell u form that's not all it was lost and you can't just sort of have whatever form here having two u forms here doesn't make sense because that would be like it fell that's not all it fell you know it's what's the new information and you can't have the m form here either that's also not okay you also get these u forms in speaker to speak interaction so in conversation so u forms are typically questions so this question homo rei this is the question i got asked whenever i did anything involving physical labor in the village and the canonical response is yes i can so here we've got a u form can you do it and then the other speaker provides the resolution to that u form here's an example from a text where they're talking about weaving and one speaker asks do we tie it now this is the white thread goes here also more generally not just in question answer pairs but just to carry a conversation forward so here's an interaction that i had in the village i was sitting around with some some guys and this other speaker came up to me and he asked me and i said i'll talk i'm sitting and then he said i'm talking so you're sitting are you and i wasn't very fluent at this point so i didn't really know what to say i just sort of continued sitting but that wasn't enough so he's then speaker a came forward and he starts offering beetle nut to the group beetle nut is the sort of canonical activity of social interaction in team war so here we have this u form it's kind of like a rhetorical question i wasn't fluent enough to resolve it so that guy then did he didn't just sort of end the conversation and walk off to the garden he kept on interacting so this is signaling i want to keep on i want to keep on talking we're not finished here's a sort of longer example about a car which came off the road in a crash we have one speaker so there will be a number of letters here and they're different speakers he's discussing what happened he forced himself and tried to reverse then he gets interrupted he was firmly stuck and there's this u form here then another speaker makes a contribution yeah he would have wanted to shrink the car make it smaller u form then another speaker makes a contribution they said the car was on the slope you said what said it was on the slope no u form but that speaker's not done it fell straight through and then they got it upright u form and then another speaker speaker b again and then suddenly this one it fell so here we have a number of changes of speaker and each change of speaker is initiated by a u form in the previous sentence the only one which isn't is the first one and that guy got interrupted so he wasn't finished but someone just interrupted so there's u form signal you know yeah let's keep talking you know this isn't finished and i did a bit of counting i've got three hours of glossed texts two and a half hours are monologues so just someone telling a story and then there's half an hour of conversations and in the monologues we have about one and a half u forms every minute and in the conversations we have two and three quarters u form every minute so this is because there's more reason to have a u form in a monologue it'll occur if you've got sort of a series of events in conversations it'll occur when you've got that and also when you're wanting to you know pass the conversational turnover to another speaker so in the discourse our u forms are incomplete they signal that it's not finished and our m forms finish it off so a u form says we're not done and then the m form says okay now we might be done so some conclusions um yeah first of all metathesis within linguistics so far people who've analyzed metathesis have mainly looked at the phonology they've you know come up with ways to explain why metathesis happens but in amirasi we also have to we have to deal with the phonology in a big way we also have to understand the morphology the syntax and the discourse if you were to sort of just look at metathesis phonologically in amirasi you'd miss a lot and i think that you know you know we've got morphological metathesis in amirasi this is sort of like a bridge between the phonology and the syntax and the discourse we've got the phonological representation of syntactic structure with a more with a morphological operation metathesis so that's metathesis within linguistics so now sort of within timor and then more generally so here are our varieties of wabmetto is amirasi if you ask a speaker of any of these varieties what language they speak they'll say that they speak wabmetto or dywan or timores they identify it as a single language and then if you dig a bit deeper yeah okay but what variety of wabmetto they'll typically give you one of these names now the borders of these different varieties are almost identical to the pre-colonial pristoms of the region the only difference is is sort of down here which represents a later migration from the north and then kusa mania which was part of the tetun speaking weharli kingdom um so we've got you know a fairly complex situation we've got some borders which represent some social reality but not linguistic reality and people care about these identities in a big way okay so here are three varieties of wabmetto amirasi rikeanor which is in the north and then fatulio in the northeast and they've all got different cloth designs okay um so this is one way to express your identity is to wear the cloth and this cloth is sort of specific to the village and the hamlet so this amirasi cloth with the blue stripes that's specific to kota otto hamlet where i was doing my field work and i imagine that the fatulio cloth is also specific to i think it was bineon i can't remember the exact hamlet bineon koa and similarly the vikano and each of these varieties has slightly different permutations on realizing their u-forms and their m-forms so in amirasi the word for three danu theon we just have simple metathesis in vikano we have metathesis with complete val assimilation tan in fatulio we have metathesis but we have a trace glide left so you get tan for the word for house in amirasi we have mid valves being raised to high after a high val in vikano that doesn't happen so we get um fatulio i don't have the data i want to know uh the word for fire which ends in a val sequence in amirasi doesn't distinguish between the u-form and the m-form in vikano the u-form is marked by that consonant insertion which in amirasi only happened before val initial and critics so aides i in fatulio we also have that consonant insertion but we also have the val assimilation which we saw in amirasi then the word for for tree or wood no difference in amirasi but now fatulio and vikano are the same so depending on which consonant is inserted you get different degrees of val assimilation so here's a table of uh what seven varieties of what meto and two other languages of of timo this is a rottanese language and then teton in the east and the u-forms and m-forms so the u-form of three is the same in all varieties but then having an m-form distinguishes what meto speakers from from the rottes speakers and teton but then within the m-forms there are differences so amirasi and aminuban are the same um foran we have obligatory val assimilation the mouse and fatulio we have val assimilation with the trace glide and then vikano and copas well they can be the same as amirasi all they they can be the same as amforan they've got a different form it would be interesting to know you know what does a vikano speaker say when they're speaking to an amirasi person and then what do they say when they're speaking to an amforan person for house most varieties have the u-form the same but then amforan and timos have consonant insertion in amforan we just have consonant insertion l umel timos we have consonant insertion and a change in the val and then for our m-form amirasi has val metathesis and val assimilation aminuban and vikano have metathesis but no val assimilation and then amforan and timos just don't have the consonant insertion for tree well amirasi and aminuban have the same u-form now vikano and fatulio they have the same form they insert a b amforan inserts a g hug timos inserts a g and assimilates the val and copas inserts a g like amforan but assimilates the val and then they've all got the same m-form thankfully the u-form again amirasi and aminuban go together but now vikano and amforan go together they both insert j without val assimilation timos is off doing its own thing with r and then fatulio and copas pattern together and then the m-form is the same so this variation is very complex and very subtle it depends on what word class you have so you've got um a numeral here which behaves differently to a noun and the phonotactic structure of the word so words which end in a vowel sequence of vowels behave differently and this stuff is it's very hard to do for speakers to do so with the cloth you know you can just borrow your friend's cloth and then identify as an amirasi person but this stuff is really hard and i've listened to speakers you know from aminuban in amirasi trying to you know do this and they can't do it right they sort of try but it's not quite right similarly when i was with ronima main amirasi informant he took me to some friends of his who spoke timos fatulio and copas afterwards we were talking about and he was saying yeah they speak differently and then he tried to copy the kind of metathesis the m-forms that they were doing but he wasn't getting it right he was sort of guessing so he was getting you know the wrong consonant insertions he knew that it was different but not how even though the patterns are regular you know it's just it's difficult okay now metathesis and parallelism so this is my syntactic metathesis we've got these two parts that make a hole and then with the discourse we've just got the mirror image now parallelism is a big deal in the timore region uh jim james fox and anthropologist at a and u has written a lot about poetic parallelism in the timore region the way you do poetry in this region is you say the same thing twice with near synonyms so here's a extract from an aminuban text in wab metho and we've got these parallelism so this land is very thin and very confined later we will eat badly and drink poorly uh so be it we will go to the wide style and the wide gate the rock of thumbesi the tree of thumbesi so to sort of uh construct poetry or really just beautiful speech you use these synonyms or near synonyms here's an example from amirasi this is a prayer um a prayer that you pray when they bring the offer tree forward in church so he starts off our loving and generous lord in the name of jesus christ the living sacrifice we give and give you these sacrifices with two different words forgive this is sort of hand to give with the hands and the other ones the more general word forgive which we received and received from your loving hand here we've got um the word for receive in a neighboring dialect so that's one way to form your parallel pair is to borrow a neighboring dialects form or to use the indonesian form it's not great but insignificance and filled with filth and dirt uh but we bring them to you lord in jesus's name amen but we've also got here with this verb we've got our semantic or lexical parallelism but we've also got morphological parallelism with our u-form and m-form and this is common in poetry speakers will do this they'll not only pair the lexical forms together but they'll also give them complementary u-forms and m-forms this is like a double parallelism there's sort of at multiple levels it's parallel now there was um yeah so parallel pairs and complementarity are a big deal in west tamar shulton orto was an anthropologist in the 70s he was writing about the political structure of rub meadow um and he provides a table of complementary and parallel pairs of which i've adapted to here so we have things like feminine masculine wife husband inside outside so each of these forms one half of a whole uh the one which is probably the biggest deal is the feminine masculine pair and this occurs in marriage so here we have a wedding taken place and this is in a village to the south of the one where i worked which is why they've got the the yellow lines rather than the blue so we've got the groom here and the bride and when two families are sort of uh uh when when a couple gets married the two families have now a different social relationship now let me get this right because it's sort of sort of complex so in the kinship system this is a quote from shulton orto the feminine masculine relationship is found to exist between two houses which are allied by affinol relationship so in law relationships the house with which the natal house has affinol relationships via its daughters is called feminine okay so if you've given your daughter in marriage or is it if you the and the one which it has such relationships via its sons is masculine okay here we are in other words the feminine house receiving a woman who is the source of life is inferior in respect to the one which is the giver of life and hence it's superior okay so if you've given your daughter in marriage then i think you're feminine in relationship to the other house but you've also got a slightly superior relationship because the woman's the source of life it's sort of a greater sacrifice to give your daughter and if you've given your son then you're masculine in relationship so there's this complementary relationship because as shulton lord hawk goes on to say but at the same time the feminine masculine relationship indicates that one cannot exist with other as life is impossible without the unity of male and female so you need both but it's not what's the word it's not equal there's a sort of there's a complementarity there precedence is the word that the anthropologists use to describe this so we have our feminine form which is sort of in precedence to our masculine form we need both okay both are one half of a whole um this is how it works in marriage but in other relationships you can also have feminine masculine relationship and it can be different so in the traditional political structure of what meto you had two two rulers one was a masculine ruler and the other was a feminine ruler both were biologically male and they had complementary roles so the masculine ruler he had the executive authority and was in charge of warfare and the feminine ruler he was in charge of maintaining ritual okay and so here it's not so much you know which one is is superior to the other you just need both for society to function properly um and you also have this with like inside and outside so the feminine ruler is located in the inside of you know conceptually in the inside of the domain um and then you have the outside which is sort of inferior so this is like my metathesis we've got these two forms which you need both of and the complementary and here the m-form doesn't occur without the u-form there the u-form doesn't occur without the m-form so we've got this parallelism in the syntax and this parallelism and a complementarity in the discourse but we've also got parallelism between the syntax and the discourse okay because of this flip-flop so what we've got is something like this this is our syntactic metathesis or m-forms and our discourse m-forms and in the syntax the m-form is complemented by the u-form which in turn is paralleled by an m-form in the discourse which is complemented by u-form and so on forever and once I apply the colors that's the end of my seminar so the end I like this because metathesis is often given as like a really introlinguistic to this get this nice like here's one example from English where people will get things mixed up and the the pervasiveness of it and this is majesty is really amazing and you mentioned that there's geographically distributed variation based on historical distribution of populations I just want to know how conscious you mentioned basically the field was like aware that these friends are speaking different yeah I just wondering how conscious of it because obviously people are very conscious of the fabric and that's very visible yeah yep yep and are there any other features that they feel so like this is kind of on the accent level of awareness like something more aware or less aware yeah I would say it's more maybe on the international level of awareness okay so people don't realize necessarily that it's happening so if you draw this to speakers' attention like they're like oh yeah they are different I didn't really thought about it and then and then yes some speakers are sort of more aware that it's different and others are less so there's varying levels of awareness but it's more on the unconscious level they can't sort of tell you oh yeah you know those people metathasize differently to us they're sort of aware that they have metathasis and that they're different from you know the Rotternees or whatever but they're not quite sure yeah yeah I thought the discourse stuff really funky that was really cool and so you've seen this happening elsewhere then discourse yeah yeah the discourse stuff no because partly we don't have very detailed descriptions of a lot of the languages and to get this sort of high level discourse stuff you really need to you know go through texts you know at a serious level the syntactic stuff there are parallels in other languages but they mark it differently so in let me bring up my map so in Rotter languages they sort of mark the same similar syntactic structures it seems but they do it by just deleting a consonant then in Fataluku which is a non-Austronesian language there's some sort of uh stressing going on as well and then Mumbai they have uh they have metathasis as well okay yeah leti yeah that's right leti and luang du as well there you get sort of you know a noun and a modifier and they sort of metathasize around one another um and there's other things going on like you know val deletion yeah it's a discourse I don't know yeah you told us some really big stuff about conversation in terms of yeah um turn taking and so yeah just thinking about narrative um m in a way um it's always going to be a u if there's a one of these clinics that yep that mark yep so do they have a distribution in like paragraph structure or they do they do so um so yes so you get these these m forms which are an m form because of an in clinic and then so what you get is you get pairs for nouns in particular because the clinic marks definiteness through 11 noun introduced and it's in the u form then when it gets repeated it's now definite so it's in the m form and you have these pairs going on particularly with the nouns so you know there was you know like in English there was a dog the dog's name you know and you have these so you have the same sort of the same sort of thing going on and then with the um um the verbs as well um yeah yeah so you can get these um m forms which are m forms you know at at one level purely because of the in clinic then pairing with the u form as well so you've got sort of multiple strategies to like get things in different orders so in fact your circles are actually within a bigger circle yeah there's circles within circles um yeah so so typically in uh narrative um your u form will proceed your m form except if you've got an in clinic and then you can flip it around right and the u form is when you've got a conversation yeah you go you go now yeah yeah yeah so that's actually um it's a so so i so you know they don't there's no obvious explanation for where they come from um in some varieties you don't get what you get a burp okay you know similar sort of articulatory features but that's a sort of a bit of a mystery to me like where they come from um yeah yeah i don't have singapore on my map i don't think which which section of the map in the launch here you have malaysian immigration broon life if not all i haven't marked yeah okay so so this so this is the fault of philippines yeah and i also palau i think palau is there so i so you know i could this way you know uh say it's not my responsibility but because you know the a and u kato gis unit made it but i can add singapore and and and palau and you know png and and thailand yeah yeah no i'm not actually a serious man yeah that was sirpin's next question is why isn't it yeah this must drive ot ot specialist crazy because you're violating faithfulness constraints a lot of i mean you're violating faithfulness a lot of advice but i actually had a question about the interdialectal variation oh yeah throw up that table with the cut nice colors what strikes me as kind of remarkable here is that your the forms which are phonologically if i if i remember you correctly the forms which are phonologically marked yep are actually the unmarked yep ones yep and the ones which are phonologically unmarked yeah like yeah yeah these ones yeah that's correct so particularly true fire yeah which is a reversal of what you've got in there so so somehow there's a mismatch between the function the the markedness functionally which you told us has to do with morphology and syntax and yeah yeah and the phonology which is yeah i mean it works neatly in amirasi but it doesn't work so yeah yeah that's right if you if you go to you know i'm fine or something yeah because you've got this sort of mismatch in amirasi between the nouns and the verbs but then you've got it for some of the nouns as well for other varieties but then do you get similar things with the verbs that there's um in the other in the other varieties that you use a nice neat thing i'm saying all you know marked verbs are going to be phonologically yeah yeah depending on the phonotactic structure i think you do i have less data on verbs because it's sort of harder to get on that you know by spending a day in a village but yeah again i i'm not an ot specialist but i would think that would actually strike them as very weird that you yeah that you're actually violating in the unmarked situation and you're not violating it in the marked situation and but then you're doing different things and it's just yeah it's sort of like um because i don't do ot either but from a sort of ot perspective it's sort of like you're putting out one fire here but that just you know creates a flood so you need to do something and sort of as you're constantly doing things it's sort of scrambling to try and get something sensible as an output yeah yeah so i sort of use some of this ot sort of language but that's right but i don't actually do the ot because you know yeah you have to have very sort of finely balanced sort of constraints yeah this this timos variety is the sort of the real weirdo because after an o you get inserted so like it's the data isn't here but like us all dog you'll get us inch which is sort of i just don't know what happened there and then they've got this R as what they've turned their J into an R it's just a mess you know historically it sort of makes sense if you sort of compare it to the unforeseen but anyway yeah yeah i was in learning this would be weird because it means in that in those varieties if you put words like tree or fire you actually never know what the unmarked form is unless you get an adjective or something yeah yeah that's right that's right then they borrow your words from people that you're other indonesian is the main one yeah so they enter into the system so so i should say in amirasi they sort of also do the metathesis there's a couple of constraints so if you borrow a consonant final word like uh i can't think of like motorbike then that won't do the metathesis in the same way because that would end up like i think in my sort of thinking because then you delete the final constant you get moth or something that's just unclear you know it's not a very high it's unclear what it is but uh you know a word final consonant final vowel final word or before any critic will even if it's phonologically unassimilated so amirasi has no D and no L but a word like duson which is an administrative level within indonesia you'll get dusne you'll get that metathesizing and even like kapaldesa which is the village chief in indonesia that comes out as kapaldesa with the the first element metathesized so they're sort of borrowing the whole phrase of a noun and another noun and then analyzing it and then putting the first one in the m form even though it doesn't it's not been phonologically assimilated and do you have any data which about lines which involve vowel alternations so you talked about cv flipping around but you know these vowel neutralizations do you get any examples of that also apply in modernity i mean maybe the data isn't there i'm not i'm not i don't i off the top of my head i can't think of any examples but yeah yeah and i and i seem to remember being slightly inconsistent for some of them so sometimes you'll get this vowel assimilation and consonant insertion other times they that is why yeah okay well i think we'll let you off the hook yeah go for a drink yeah