 Today, I want to say here, and hello to those of you who are on Zoom as well. I'm Joey Dockstrain, I'm a postdoctoral and investment steer, and I'm just here to present our speaker for today's seminar, Charlotte Hemmings. Charlotte did her PhD here at SOAS, and then since then has been doing a few projects at Oxford that we had on Leverhulme. As an early career fellowship at Oxford, to continue her work, which is with language with OsaraWalk, which is the Borneo side of Malaysia, working across the region languages, and now is on another project in collaboration with Leverhulme, who was my supervisor at Marca in the Indian language, that started at the University, working on this isolate language in Ghana, and let you tell us about film Indonesia, but a very different part of Indonesia. So, Charlotte, if you're joining us, we look forward to hearing about this very interesting project. OK, yes. Thank you very much for having me. Thank you for the invitation. It's very nice to be back at SOAS and to be online as well, I suppose. Yes. So today, if I could get the slides to work, what do I press? OK, so in the seminar today, I'm going to attempt to discuss the value and the limitations of using historical resources in the context of language documentation and description, using some case studies of the project that I'm currently working on to document the Ingana language of Indonesia. Ingana is in a relatively fortunate position for a minority language in that it has a rich history of documentation. There are word lists and some textual materials going back to the 19th century, as well as a systematic documentation conducted by the German scholar Hans Keiler in the 1930s. And what I'm going to do in the rest of the presentation is to give you some examples of how we've used these materials, firstly, to help in the description of contemporary Ingana morphosyntax, and secondly, to analyse parts of historical change. And in so doing, I'm going to argue that the historical materials can be of great value to language documentation and more broadly to linguistic theory and typology as well. So how we're going to get there is we'll start with some background on the Ingana language and introduce the historical materials that we've been using, then look at some examples of how these helped in discussing contemporary Ingana morphosyntax, look at a case study of historical change and then come to some conclusions. And I should say you're very welcome to ask questions at any point if you would like to. OK. So Ingana is spoken by approximately one thousand five hundred speakers on Ingana Island, which is situated off the south coast of Sumatra in Indonesia, roughly where the red pointer is here. It may not be immediately apparent from this map, but it's actually pretty isolated. It's around 10 hours by ferry to the nearest coastal town of Benkulu and even further to the closest island in the Bari Island chain that Ingana is part of. Perhaps because of this geographically isolated position, but mainly due to the relatively unsurprisingly low cognate percentage with proto-Austronesian, there has been some historical debate as to whether this is an Austronesian language or not. Today, most people would probably agree that it is an Austronesian language on the basis that there is cognate morphology with proto-Austronesian, proto-Malaeo Polynesian, but there's still ongoing debate as to whether the language subgroups with the other Bari Island languages in the chain here and the Bari languages of Sumatra or whether it forms a primary subgroup of Malaya Polynesian. If we zoom in on the island, you can see that today, there are six villages where Ingana is spoken, all along the north coast of the island. We know from the historical records that previously there were other settlements along the south coast of the island as well. We also have an anecdotal mention of dialect variation. By the time we get to Kayla's work in the 1930s, he says there's already been quite significant dialect levelling as a result of pretty drastic decline in population in the 1800s and the resettlements of all the populations along the north coast where you're left both to the Indian Ocean. Today, we don't know at this stage of the project, we don't know an awful lot about the dialect situation in Ingana, but what we can see is that there's different levels of effects of contact with Indonesian. So although the language is considered endangered across the island, the speakers are increasingly shifting to the national language, which is Indonesian. The language tends to be more or most vital in the central villages, such as Meil, where there's a larger proportion of Ingano speakers. In the northern and southern villages, there's a much higher percentage of non-Ingano speaking populations. And this has led to a federated process of language shift. I suppose so. Yeah, maybe we can talk about that a little bit more at the end. It's a very interesting question. So as I mentioned, there's been quite a wealth of previous documentation on this language. So here are just a kind of selection of some of the word lists that we have available to us, and you can see some of them were reasonably expensive, I guess. And they're the full list of all the word lists that we know of on the Ingana project website just to give you a sense of what it looks, what the records look like. This is the Walland list, in fact, the only list that mentions here, you can see a kind of high language and a low language variety, which may be familiar to you from Javanese or Balinese. No one else mentions this. There's no kind of in our work on Ingana, there's no high or low variety today. Make of that what you will. What is interesting is that the contemporary Ingano forms tend to correspond to what was Walland's low language variety. Here's another example of a list. This is HealthFish and Peters and some years later, HealthFish published an updated version, I guess, as you may already be able to see, there are some kind of differences in transcription, even between the same authors and within the same text. So here, if you see Baka Kaha and Baka Uwe, these probably both contain the form Baka, meaning I, but it seems to be felt differently, exactly what that means we have to guess, I suppose. The HealthFish 1916 text also has some text in it, which is really nice to have these materials going back all that way. And they're published like this. So you have basically the Ingana word with a line by line, word by word, Malay translation, and then at the end of full translation into Dutch. So we have these materials. That's great, but there are several challenges if you want to actually use them. Firstly, there's a kind of accessibility challenge. So the materials are translated into Dutch and or maybe Malay in some cases, but certainly not in all cases. They also follow the conventions of writing Dutch in the 19th century. So you have to be aware of what those are. For example, the sound, the digraph or the letter, combination of letters, O, E, you're nodding. So perhaps, you know, this already corresponds to the sound. So, you know, you have to know these sorts of things. And there's also a question of comparability. So to what extent do the different kind of work, if we find variation in the word, to what extent is this different dialect variation? To what extent is it different practices of transcription? We have to somehow use an educated guess to make that judgment. And as far as I'm aware, we don't have the original manuscript. So we don't really know what happened between data collection and publication of these resources. The second main historical reference that we have available to us is the documentation by Hans Taylor. And this involves or includes a sketch grammar that was published in 1940. Sorry, I should say he spent seven months on Angana Island in the 1930s and the later published various resources, including a sketch grammar, a text collection of a variety of different tax stories, including great stories and kind of descriptions of cultural practices and a dictionary that was actually published later by one of his or several of his former students based on the kind of notes that he'd put together over his lifetime. Again, I'm just flashing up what they look like on the screen so you can see here you've got kind of sentences with the German translation. You've got entire texts followed by the German translation where roughly sentences correspond to sentences, but maybe not always. And the things that I've circled are just to show you that again, even between the resources collected by the same author, there are differences in transcription, which may or may not reflect kind of changing analyses of the same phenomena or who knows. So again, we have the same sort of challenges really. We have this accessibility at the moment you have to be able to speak German in order to access these materials. There's no glossing, so it's not readily available for doing the kind of, I guess, quantitative studies or qualitative study that we might want to do. And there's this question of comparability, particularly with the earlier resources collected by Dutch administrators. To what extent do differences reflect just different transcription practices or change or dialect variation? And again, we don't have the original manuscripts, so we really don't know what happened, how the data collection took place and what happened between the data collection and the data publication. So what we've tried to do then in order to limit as far as possible these challenges of accessibility and comparability are to kind of work with these materials and make them as accessible and comparable as they can be. So we've had the materials translated from Dutch and German and to English and Indonesian. And as part of the project, we're going through and trying to produce a kind of flex database with interlinear glossing for these materials. This is a process that's ongoing and there are several kind of decisions that have to be made about where we have real differences and where we have just different transcription practices. But we're ending up with something that looks a little bit like this, where you can see the texts that were not so accessible become hopefully a little bit more accessible for speakers today. So after Kayla, several other people, have worked on Angano and it became clear from more recent publications and from our own initial work or I should say Mary and Ryan's initial work on Angano, that contemporary Angano has undergone quite dramatic changes compared with the variety of documentaries by Helper and Kayla. And you can see a little bit in these examples, the contemporary form versus the older forms where here we might think of the difference between Lofor and Dofor probably reflecting dialect variation because we know that D and L tend to alternate with one another. This moor is probably just a difference in transcription practice. So I understand that it was common in Dutch publications of the time to use the omelette to indicate two vowels that were separated by a glottal stock rather than a long vowel. Here it's not entirely clear whether this represents a difference or just a difference in transcription. And because you'll see that sometimes Helper uses the symbol omelette to represent the centralized vowel that Kayla, at least in the dictionary represents with the schwa symbol. Since these are related, that the root C is in both of these forms, one would expect it might be the same vowel and certainly it is in Kayla, but it's not clear. It could be that there was a difference in the vowel quality in the variety that Helper documented. So that just as an illustrational case of where it isn't clear what to do with it. But what is clear is that the contemporary forms have systematically lost the final vowel compared with the older recorded materials. So these changes together with the sociolinguistic context that I described not very well at the beginning of the talk provided the motivation for our ongoing AHRC funded documentation project. So some nice pictures of everyone doing some nice documentation in the pre-pandemic world. So you can see, Joey already mentioned my colleagues, Mary and Wyand on, I think one of their first visits to Engano working with local speakers, with community members, with local researchers and the training community members to help in the documentation project too. The overall aims of the project are to collect an archival documentary corpus of contemporary Engano to produce a grammar based on the database of these glossed texts that we'll collect to produce teaching materials again, based on our understanding of the grammar of the language for use in the community and then to assess the position of Engano in the Austronesian family given this long-standing debate over subgrouping and typology. So data collection had begun before the pandemic started. After the pandemic, we've moved to this kind of newer model of documentation where everything happens online, totally unsaged photo of us all smiling happily during every session. And we use these sessions to go through the materials that had been collected, listen again to the text, and think about what the glossing would be. And as a result of this process, we're building a corpus of glossed texts that we hope is broadly comparable to the K-LA material. So we're building these kind of parallel corpus together. So to summarize what was very long background, what we have at the moment is an ongoing documentation project where we want to document contemporary Engano and consider how it relates to the wider Austronesian family. We're lucky in that Engano has a long history of documentation, including the 19th century materials and the documentation of Hans Kehler. And in what follows, what I want to illustrate is how we use these older materials to help us in the aims of the ongoing project, sort of as a vehicle for exploring the value of historical or legacy materials more generally. Okay, so let's start with some potentially puzzling data that became sort of apparent to us as we started to work with contemporary Engano. Okay, so let's say you have a word like back, I. We mentioned this earlier. Let's say you want to make it my I. Okay, you could say back, ooh, ooh being the pronoun for I, me. But you could also use the suffix form back up. For my I, back come, for your I, your back be for his I. I'd like it some more. Okay, the word for how? Well, if you ask a younger people like Enga, then you assume, so again, you can say, ooh, my house. But then if you want to use the suffix form, that becomes your back, your bam, your day. Another one, your forehead. And do the same thing, your ooh, my head. If you want to use the suffix form, your room, your day. So now different vowels. And then there are some that are kind of just a bit funny, I guess. So yeah, it's hand, if you ask the word for hand, you can say, yeah, ooh. But once you use the suffix form, it becomes up, up, but apparently, yeah, day. So what's going on and how might we interpret this sort of messy or puzzling pattern? Well, one of the things that we could do is we could look at the pevedorical resources to see how possession was marked, or how Kaila thought, how Kaila thought possession was marked. According to the grammar at least, possession used to be marked via pronominal suffixes that regularly attached to the stem. So we can look at this with the example of house, which in the Kaila documentation was with the noun marker A at the front. So all nouns had case marking in Kaila's variety, which I'll henceforth call Old Angano just for one to the best of word. So you can see A-U-Ba regularly takes the suffix to indicate the person and number of the possessor. A-U-Ba, A-U-Ba, boo, et cetera. Kaila also tells us that suffixation triggers stress shift. So it's something like A-U-Ba, boo. A-U-Ba, au, something like that. Whereas any fertilization with the form D-A, die and do apparently does not. And that's sort of important to bear that in mind. Okay, so as an initial stage, this gives us a sense of what we might see if it's that. We might expect there to be suffixful forms for all of these range of possessors that we can go and we can work with speakers to see if these forms exist. And sure enough, they do when we went back from the text with all the speakers, we were able to put together a kind of paradigm of possessive markers that is maintained in contemporary Angano. But what you can see is that the kind of connection between the root and the suffix form isn't really as transparent as it was in all the Angano. So in exactly those person and number combinations where Kaila said there was stress shift, we see this vowel kind of reappearing. So you're bum, you're back, you're bum, you're back, et cetera. And in the cases where Kaila said it was Angodicization, you see no vowel. And that's the same pattern. If we look at all of these nouns that I mentioned, except that the vowel is a different vowel. So how do we know which vowel it's going to be? What do we do with this? Again, looking at the historical materials that the lexical materials gives us a kind of explanation for these patterns because you'll remember when we looked briefly before, it seemed like the contemporary Angano forms had systematically lost the final vowel. And so this is the vowel that is essentially resurfacing as it were in the suffix form. So looking at these, comparing the lexical items gives us a sense of the regular sound changes that have occurred between old Angano and contemporary Angaman. The most obvious one that we've already noted is that the final vowel is regularly lost, but there are certain other processes. So for example, voiced consonants are often le-knited. So the verb is realized as ma by younger speakers in particular, and de is realized as r by younger speakers. So hence, there's alternation between yub and yum for younger speakers, but it's that underlying consonant that is realized under suffix asian. Another change is that, oh, Blitzen is sometimes realized as schwa, which we represent with the symbol, hence why we get a put rather than a pop. And we see a process of what we might call palatalization of vowel initial roots. We assume that following this noun marker a, a glide was added. So a yuba becomes a yuba, I guess to make it easier to pronounce. At some point, it became optional to have the case markers and this was lost, but the glide was reinterpreted as part of the root. So relatively systematically, we tend to find a glide at the beginning of nouns now, hence yum, yub and yup. So that kind of gives us an explanation for what's happening. Basically where we have these suffixes added and they triggered stress shift, this final vowel is stressed and so it's resurfaced, it resurfaces or it's realized in this part of the possessive paradigm, but where there was no stress shift and the vowel there would be unstressed, it's not realized in the paradigm anymore. And this gives us exactly the vowel that we would expect from the caliform is the one that resurfaces in the possessive paradigm. From our initial data set, it was also clear that the possessive suffixes were not the only strategy for marking possession. You'll remember when we talked about my house, we said we could either say yuba or yubu. The second strategy with the free pronoun could be considered a contact induced change since it represents a calc from Indonesian. This strategy that you see there on the right, uma aku or uma saya. And this strategy most likely developed in response to the phonological changes that affected the regularity of the possessive paradigm in Old Angano. So what looked like quite a puzzling data set is much more explicable when you have reference to the historical materials. And in fact, these kind of processes of vowel deletion of what we may call resurfacing under suffixation and the loss of non-stressed vowels and penultimate syllables are general processes that affect contemporary Angano across the morphology. So knowing that these, knowing that the final vowel resurfacing indicated the presence of some sort of suffix can be really helpful in terms of analysing the text that we are collecting now. So again, if you look at kaha, which is an intransitive variant of like versus kappi, some transitive variant of like, maybe the connection between the two forms is somewhat opaque. And if we compare with the older form school, sorry, in Kayla, at least again, this was regularly a process of root plus affixation. So ki plus a-ha-pi would have given you kaha-pi becomes kaha regularly by the loss of the final vowel. And kappi would have been ki plus a-ha-pi plus an applicative suffix he. You lose the final vowel and then you lose the penultimate non-stressed vowel here and you end up with kappi. Again, hek and heku, you might think, oh, well, it's a process where you add a u on the end and it's derived something that different nominalizations have different vowels at the end. And again, this is because this is like, what was a regular suffixation process in all Denganos, but now you're losing that final suffix and the only indication of morphological process is the presence of the final vowel. Of course, these may be most likely are lexicalized for speakers, I'm not necessarily assuming that these are doing these kind of processes online, but at least for us, going through and glossing the text, it gives us a kind of metric for looking at them. If you can see the final vowel that was there in Kayla, probably there's some kind of suffix that's been added and you can think about what that may or may not have done. Okay, so that was the nominal morphology and that was relatively simple compared with the verbal morphology. So now if I have time, hopefully, I'm going to run you through some of the complexities or verbal morphology too and look at the more prefixes, I suppose, and think about again, how the historical materials can help us just simply in analyzing the data that we're presented with. So if you ask, if you have a verb root like pu, which means run, you can derive various different forms and we tend to think of it now as kind of three basic forms, the key form, the boo form, and the bear form. In the key form, you attach key to the root and then you express your subject via a free pronoun, which incidentally for the third person is also key. If you have the boo form, you attach one set of subject agreement markers and the third singular is ka and in the bear form, you just have the bear root and then you attach a different set of subject agreement markers for the third person is e. So far, so good. The difficulty with establishing which kind of form you're dealing with on any particular occasion is that there isn't always a distinction between the three pronouns and the different agreement sets. So for the first singular, for example, the free pronoun, the first set of agreement markers and the second set of agreement markers are the same. And then there are some verbs that don't take boo in the context where we would expect them to take boo. So then the root is kind of the same in both of these contexts. That already makes it quite difficult to know in any given context what kind of verb you're, what kind of analysis you want to give to your verbal morphology. It's made potentially more complicated by the fact that there are several derivational ethics that can attach to these roots that begin with an a. And that makes distinguishing in particular between key and ka quite difficult because if you have a root that begins with a vowel, then you will get the ke form of key. So for example, if you have this form kabari and it came up in a text that looks quite like this boo form kabupu, then you could potentially analyze that, I suppose, as the ka agreement plus a b for boo and then a root kari, but it seems like this is probably more likely key plus there's more theme that we'll talk about in a second and plus pari. That's basically complicated is the point that I wanted to make. And so we're faced with kind of these texts where we're not entirely sure whether we're hearing. In general, you're trying to transcribe these texts that you don't know whether you're hearing them correctly and you also don't know how to analyze them. What can you do? Well, you can use the historical materials to help guide you as to what might be the organizing principles of the grammar. So based again on the trailer grammar steps and our analysis of the text materials, this is what gives us this idea that verbs generally split into three forms. They can form the boo form for the bear form. What it, it appears that key verbs mainly occur in relative clauses and they seem to be an innovative structure that is found in Angano and also in Nias, one of the other barrier island languages, but not generally found in Austronesian languages of the region. Boo verbs tend to occur with a set of subject agreement markers and bear verbs occur with another set of subject agreement markers and those are just kind of parallel examples to the ones that we saw from contemporary Angano a second ago. The grammar schedule also gives us the forms of the agreement prefixes. So you can see they differ in the second and the third person but not in the first person. And they give, and our textual analysis gives us a sense of where the different forms occur. That gives us a way of distinguishing between a boo context where the verb simply doesn't have a boo and a bear context where we really would never expect there to be a boo. And so what we find is that boo verbs tend to occur in main clauses where they're typically verb initial and in embedded verbal constructions following various auxiliaries including whole heap of high and the negative imperative form where they occur without the agreement markers. Bear verbs tend to occur following negation with subject agreement markers of the fact two in imperative and in clause chains where they often combine with another derivational prefix aber. So that again gives us a model of something that we could go and look for that might not have been immediately apparent if we had only access to the text. We might not have thought even to look for subject agreement markers but sure enough working through the context of negation and these contexts where we expect these forms to occur with some older speakers who were able to elicit this paradigm. I didn't really mention this before but it looks like for some younger speakers particularly in this context of negation they're moving towards the three pronouns followed by the negation followed by a default agreement marker and then the verb root. And I believe that that is also attested elsewhere in languages of Indonesia that have developed these agreement marking systems. So using all this is all by way really of showing that you can use the analysis of the old materials to help you decide in a particular context where you have a kind of tricky form what analysis you're going to give it. So here this is from elicitation I suppose. We know that it's not verb initial there's a subject in the initial position so we're pretty sure that this is where a context where we would expect key and so we can give it this sort of analysis and then comparison is another case of the same word but it occurs now in a different context where we would expect some sort of agreeing form. I wanted to give you the same word so that you would have a kind of comparison but I'm actually not really 100% sure if this is the right analysis but at least it should not be key it should be the agreement form pair whether or not what follows it is correct that remains to be seen. Perhaps this hasn't been very convincing but what I wanted to show you was that looking at the historical materials can give us a whole load of things that really help us when we're faced with morphologically complex language that we don't necessarily know what to do with. It can give us a guide for elicitation so it told us that we might expect to find possessive the paradigm of possessive suffixes and that we might expect such a agreement market. Yeah. It sounds like one study is enough and it sounds like if you're talking about historical materials it could be other kind of gaps in that history that could help you in terms of materials when you're just having fun with it. You mean looking at the older looking at other materials in addition to Kayla? Yeah, I think that would be the next stage for the project really to go back to the other places where we have textual materials and see if that gives us some extra answers. Yeah. So it gives us a guide anyway for elicitation. It can give us an explanation for the patterns that we find and it can give us a kind of method for how we might go about our analysis of the contemporary text. And in my opinion, at least it's quite difficult to imagine how we might have tackled this, particularly given the context of not being able to go to the field and working online over Zoom. Maybe some of this would have been different if we had done more traditional fieldwork. But it's difficult for me to imagine how we would have started analysing the purpose without access to these materials. Okay. So if I have time, then the next part of the talk is to look at how having not only lexical materials but also textual materials from various time points can allow us to analyse paths of change and the impact that this can have. So we've been using the Parallel Text Corporate to analyse various aspects of change. So we've looked at, for example, at the increase in lexical borrowings and phenomena associated with language contact. We've also explored potentially contact-induced changes such as the development of the new possession strategy that I mentioned before and shift and word order. So I mentioned that in all the Ingano-Druklauses words generally verb and noun, and in contemporary Ingano, it seems that they're moving towards an SVO structure, which doesn't help us when we're going back to how we analyse, how we decide what form we have, of which particular verb. But today, what I want to focus on for the rest of the talk is a case study of a particular construction as an illustration of how looking at change can have important implications for theory and or typology. By way of background, for those of you who are not familiar with the Austronesian languages, they are known for their symmetrical alternations. And these are alternations in the verbal morphology that affect the mapping of arguments to functions. Kind of canonical example comes from a language like Togalog where you can come to the verb, which takes the different marker, the um and tx or the in and tx, and this corresponds to a different mapping between arguments and functions and the actor that's the subject and takes the nominative case. In the undergoer voice, it's the undergoer that's the subject and takes the nominative case and the other argument takes the genitive case. So we have different kind of verbal morphology, different case mapping patterns, and this corresponds to different mapping between arguments and functions, but importantly, two transitive clause types. So multiple types of transitive clause. In many language, the subject has, there are various properties that are restricted to the subject. So there can be syntactic reasons why you might choose one voice over another. So if we think about actor voice in particular, you can only extract or relativize on an actor. Um, yeah, sorry, if you have, if you, if you want to say, you know, the man who's read something, then you can do that. If you have an actor voice clause and you're relativizing on the actor, but you can't relativize on an undergoer if the relative clause is in the actor voice. So you can only kind of relativize on what would be the subject of that embedded clause. Um, and there are also kind of semantic reasons or functional reasons why you might choose one voice over another. So it's well documented, but in the more conservative varieties like Tagalog, actor voice has semantic properties associated with, with antipathics. In particular, um, the, uh, undergoer argument tends to be indefinite, non-specific or less affected. Um, and you can see that from the example here in the actor voice, kind of construed as hitting at, uh, Jose, rather than hitting him and making sure that he's definitely hit. All of this leads to a kind of problem in terms of, uh, addressing, assessing alignments in these languages because as you will know, traditionally alignment is identified by comparing a transitive clause with an intransitive clause and seeing whether the single argument of the intransitive clause behaves like, um, the A or the P argument of the transitive clause. If you have more than one transitive clause, which transitive clause do you compare it with? If you compare with actor voice, it looks like accusative alignment. If you compare with undergoer voice, it looks like negative alignment. Um, so elsewhere, following various people, uh, we've argued that the best way to go about identifying alignments in magical voice languages is to use what we might call a functional-marketness approach. Essentially, what you want to do is to find the least marked or the most prototypical transitive clause type and use that to identify alignment. So if your, uh, undergoer voice clause is the most kind of canonical, or the least marked transitive clause in terms of its semantic and discourse properties, then alignment is ergative. And if actor voice is the least marked, then alignment will be accusative. If you adopt this kind of model of alignments in Australian languages, then we can, uh, we can make a case for more innovative languages, particularly those where actor voice is marked by a homorganic nasal, rather than the sperm infix that we throw into garlic, have undergone some sort of alignment shift via the reanalysis of actor voice as kind of from something more anti-passive-like to something more active-like. Hold that in your head as background. So the reason I provide this as background is because although it looks nothing, not doesn't look that much like it, the AH construction in Nangano appears to be cognate with this homorganic nasal that we find in many Australian languages as a marker or as the main marker of actor voice. And we claim this on the basis there is a correspondence between AH in Nangano and Proto-Malaeo-Polynesian N and much like the homorganic nasal in many Austronesian languages, this prefix appears to trigger changes to the initial consonant of the root that it attaches to. Among other things, if you have a root that begins with P, like Paris in that Cabaret example that we had right at the beginning, then under affixation with the in this AH construction, you will end up with B or M depending on whether the root contains nasal elements. If you have a root that begins with K, it will change to D or N, again, depending on whether the root has nasal elements. And this can be understood as a process of nasal substitution, the same thing that we find in lots of Austronesian languages combined with the Nangano specific property of having word-level nasality. So, basically, you couldn't have a nasal on the front of a word that didn't contain any other nasal element. So, now we can use our textual materials to look at how this construction is used in real data. And what we see is that it behaves more or less like a prototypical antipathic, so not really like an active voice, not like active voice at all, like a morphosynaptic antipathic, because when we have the AH prefix, we see that the P is demoted. And you can see that nice and clearly in Old Nangano because it had the system of case marking. The A marker that we've already encountered was used for direct argument, that is subject and object. And the OO prefix here is used for oblique argument. In the context of this AH construction, or we might call antipathic, there's sometimes also the generalized preposition, e.e. or you can have either just the oblique case marker or the preposition as well. We have some more evidence that it really is syntactically intransitive because Nangano has this property that in subordinate clauses beginning with R, keep with this kind of critic R, you tend to find the boo form of an intransitive verb, but the bear form of a transitive verb. And whenever you have the antipathic form, it's always preceded by boo, in this case, the N-elomorph. So it really is syntactically intransitive and it also has kind of semantic and discursive properties associated with antipathics too. In particular, you can see that often the P-argument of this AH construction is indefinite or generic compared with P-argument of the construction that just uses the bear root. So these two examples are illustrative. The first one comes from a folk story about a woman who gets kind of spirited away by spirits and then she comes back to see her family and brings them food regularly. And what's important in the context of that story is the action of bringing food. It's not important what food she brings, just generic food in general. In the second example, and this is a story all about preparing for a major festival, and prior to this particular example the speaker had mentioned particular types of food that had been collected and bought to a location, then said something else and then come back to it and says then they take all of the items of food. So it's referring to specific items of food that had been previously mentioned in the discourse. In keeping with the discourse profile of antipathics, these constructions are also relatively infrequent in naturalistic discourse. We had one story that we collected contemporary retelling for. In the old, the original version, there were only two instances of this construction in the whole text, and just to put that in context, in the almost 40,000-word corpus so far, or not so far, that's the corpus. There are only 507 instances of that morpheme and then a couple more other constructions that we might consider to be functionally equivalent. And probably about half of these are used in action nominalizations rather than in these verbal antipathic-like constructions. So it's pretty infrequent. If we compare this with what we see in contemporary Angano in the text and in elicitation, we can see that the construction is no longer behaving like an antipathic. It seems that you have a core argument regardless of whether you have the AH morpheme or not. And we can tell that these really are kind of core arguments despite the fact that the case system seems to have been lost because obliques in contemporary Angano are systematically marked with this generalized preposition. Oh, that probably comes from EPEL. And that's not optional. You can't have that preposition in construction containing the AH prefix, but what you can have is this kind of optional direct marker occurring. So by all evidence, it looks like really just a normal, just another transitive construction. It seems like there's a bit of an association with maybe imperfective aspect, which may be kind of, which may be a hangover from an approvious antipathic-like function. And we know that often antipathics are associated with imperfective reading. But you can definitely have a very highly affected undergoer argument, such as a pronoun, for example, or very definite and highly affected. Definite because it's a pronoun and highly affected because it's been pinched. It's slightly more frequent, although maybe arguably not that much more frequent in the contemporary retelling of the story, there were 13 instances over the two. So that's roughly 4% of verbal clauses, not that frequent. But what was interesting and what we'd like to look at in more detail as we kind of build the corpus more in a thesis that our, that Bindi Vidhaya wrote into 2018, we saw that AH constructions were used in, more than half of the example sentences that were given as translations for Indonesian members containing this nasal, homo-organic nasal form. So perhaps there is some connection between four speakers between these two forms that were historically cognate. So more to be done. And the next step I think would be to go back and look at the older texts and see how this construction was used there and to continue to analyze this in the contemporary corpus. But as a kind of mid-summery here, lots, a number of Western Australian languages have been claimed to undergo an alignment shift as a result of this re-analysis of something from anti-passive like to active like. In Angano, we see a very similar process of re-analysis. An anti-passive doesn't look like an anti-passive anymore, but what about alignment? Well, neither old Angano nor contemporary Angano appear to have a symmetrical voice system. Instead, the morphology that was previously associated with voice, that is the choice between boo, cognate with this womb infix that we saw in Tagalog and bear forms potentially from an embedded undergo voice appear to mark a kind of tense aspect mood distinction instead. So what once marked symmetrical voice now seems to mark the distinction between what we might call realist and irrealist in that this construction is largely limited to following negation and imperatives. There's also no extraction restriction. So that kind of active pivot function of the active voice doesn't isn't linked to boo versus bear anymore. So if you imagine Tagalog, if you wanted to relativise on the actor, you'd have to have your active voice form. And if you wanted to relativise on the undergoing, you'd need the undergo voice form in your relative clause. That's not the case in Angano. It's not that you need boo if you relativise on the actor or bear if you relativise on the undergover. Instead, you use this new key construction in all relative clauses. So no evidence of a symmetrical voice system. And instead, lots of morphosyntactic evidence pointing towards accusative alignment in both the right holding on and contemporary Angano, including the fact that the verb index is F and A, both of those different agreement sets that I mentioned. And the fact that F and A seem to function as a pivot in terms of relativisation, because when you relativise on F and A, you use this key form in the relative clause. But if you relativise on O, at least when it's a full NP, then you tend to use a nominalisation construction step. So this would be something like the woman... woman that is the one who was seen by our father. So consequently, Angano appears to be accusatively aligned in both old Angano and contemporary Angano. But it does still seem to have undergone this kind of reanalysis of what looked like an antipassif as something more active-like. And this suggests that this process of reanalysis can occur independently of alignment shift. It's probably the result of the particular developments in the prehistory of Angano that led to the various functions of active voice, the magical voice alternation, the act of pivot marking and the object of motion semantics being split between three different constructions, the VU form, the key form and the AH construction. But either way, it reveals a really interesting set of possible developments in Austronesian voice morphology and provides further support for the idea that antipassifs are particularly amenable to historical change, independently of alignment shift and most unusually in the Austronesian context, independently of symmetrical voice. So what I hope to have shown you then is that the historical material has not only helped in kind of guiding the documentation and the building of the documentary corpus, but also enable us to analyze parts of historical change, which isn't always possible with under-documented languages but can have important implications for our wider understanding of linguistic theory and typology. Hence, they make up an important part of achieving the kind of goals, the wider goals of the project to assess the place of Angano in the Austronesian family. So that's it for me. What I want to conclude by saying is that Angano has a rich history of documentation and is relatively fortunate, I suppose, in the relatively fortunate position of having parallel text corpora from the 19th, 20th and 21st century. The historical resources are limited in ways that we've discussed and present challenges in terms of their accessibility and comparability, especially since we don't know what the data collection process and publication process really involved. And there's a very interesting paper that you may or may not have read, which looked at exactly the difference between kind of published older records and going back to the original sound recordings and the original manuscripts and the kind of editing process, I suppose. So I do read that paper. However, they give us an important window into what is, I think, quite a morphologically complex language providing an explanation for puzzling synchronic patterns and a method for analysing new texts, as well as analysing diachronic change with potentially important implications for Austronesian and for typology. Consequently, I would argue that historical materials can be very useful in the context of language documentation and description and that making these materials accessible and comparable for future research can and should be an important part of producing transparent and contextualised records of a language. So that's it. Just left a thank to the community, the various people working in the project have contributed in various different ways and the funders and you as well. Thank you. Thank you very much, Charlotte. We've got a lot of time as we want to stay around and ask questions for. We can stay and ask questions. So we'll start people in the room, but if you're online and have a question, you can write in the chat where your question is or you can use the raise hand function or just make a note in the chat that you want to ask a question and we'll get to you. So now for here, do anyone here have a question or constructive comments for Charlotte? Chris. Yep. Thank you so much, Charlotte. Really nice to see you back. Thank you. So I'm very sorry at the beginning when I'm trying to go forward. It's rather important to start with you. Okay. So I mean, I thought it was an absolutely fantastic talk. So I really hope this question will come across well. So speaking as, you know, I really can't recall myself feel linguist. I just feel a bit confused by the way you present this idea of using the historical materials as much as possible. You sort of present that as possibly surprising, but... Edwin, isn't it sort of obvious that you would do that? Is that what you were saying? From my perspective, which is, you know, far less, you know, with no expertise on this sort of documentation. So I mean, you know, please tell me what I'm missing and what's wrong about this. But I would just, I would assume, I would sort of say, okay, well, this is a language which is poorly documented, but it's documented. Documentation has all kinds of flaws. And one flaw is its own. And therefore we can expect some amount of change. But there'll be other things like poor analysis or whatever. Yeah. So just to summarize the question for those online, Chris is wondering, sorry, why do we have to argue for using historical records for this kind of work in the workplace? Isn't it obvious? Yeah, I mean, I guess you're right. And maybe you don't need to argue. Maybe, maybe what I wanted to say was how you could use them rather than that you should use them. I think sometimes people can say, oh, well, you know, we can't, you can't do anything with these materials. So we'll just focus on what we have, the data that we have available to us. And I guess the point that I wanted to make was, although it's taken a really long time to make these materials accessible in some way, I think it has been really worth it because it's enabled us to do all these things that we might not have done otherwise. But maybe it is, maybe it is already obvious. Yeah, I wanted to tackle that though. It does take a lot of effort. I mean, so the question is, is it worth the effort, but also do funders care about this kind of thing? It's not very sexy for funders to go like reanalog old documents. So how do you argue for why is this essential to this bigger picture of documenting languages? I understand. Another question. Yeah. I think one of the main points was that we don't normally get this much historical information from, so why do you think it is because it's, you know, positive. That is a good question. I don't really know. So just to review the question, why is the Ghana so well documented compared to a lot of minority languages that haven't been documented? Maybe historical accident. I know that Kayla did do quite a lot of research in that area in general. And I think perhaps he wants to do something else and then ended up on a Ghana Island. So maybe that does historical accident that he happened to work on that particular language. I don't know if my experience working in Sarawak was that they're really not very many historical records at all. And what existed was kind of word lists of maybe 30 words or not very much at all. So I found it very unusual to have all of these historical materials available to me. But it may not be actually that unusual, but that I don't know that would be an empirical question. Another question here. Yeah. So it's more about in Ghana itself. But you mentioned in contemporary in Ghana, their form is generally here in Europe. And you mentioned negation and comparatives. So I was just wondering if it's also usually the same kind of verticals or certain conditions as well. I don't know about conditional. That's a good question. How would you do that? What I know is that it's used in those contexts. So following negation will always get either that default E or one of the other green markers plus the bare form. And if you want to form an imperative, there are actually two ways of doing it. You just use the bare form. I mean, that's normal, I guess. Or you can use the blue form. You know, a second singular or second plural agreement form. So there are two kind of strategies for marketing imperative. Really. And then there will be clause changing constructions where you get the first verb is in the blue form. And then subsequent forms are the bare form with often with the verb abba in front of it. That I don't think you can really call it, but maybe you could think of it as an embedded construction. So when they're just, they're just one verb. And that verb is either blue or bare. Then it is kind of a realist construction, but it's not quite as simple as I maybe made it sound. And it seems to me like you had, you had these forms that originally marked a voice alternation. And then they are inherited in some way. So you have this choice between boo and bear. And it's not marketing voice anymore because it doesn't, the choice of the bare form definitely doesn't mean that the undergoes the subject. But they just kind of turn out in different contexts that may be somehow linked to an earlier function of voice marketing. Yeah. I'm just chucking a tiny comment there. Go on. There's been paper by Roger Lass how to do things with jump. Have you come across that? No. It's also an exaptation when more themes lose their original function and they tend to be exacted or whatever the word is to a new kind. Sounds like that's what's written on it. That sounds like my analysis of what's going on here, whether or not that is true. I don't know how to do things with jump. Okay. I will make a note. I did not bring a pen. So I'll have to use my phone. Well, that's a question. I mean, on this, on this stage already, I think because it might not be that it's done, but I was still a little bit curious whether you can push further what it is that's happening. But I think you have shown very nicely that it's not creative change. It's not alignment. Yeah. But then, but then it may well be something different. And the reason I'm asking is, and we had this discussion before that other postination language is tantalizingly closed about the language. It's like post-exactuals, but the language is actually bigger. It's the various issues. So we've been struggling for a long time, but with the causes of the bigger tips. And then, and then the standard analysis is what that's syntactic. So, you know, and that's a little bit like the line machine, but then there are lots of functions where you have the morphology, but it's not syntactic. And now I get what the people are thinking that too, that the syntax is a function of something else. There's something deeper going on, which has to do with that semantic or, you know, a grammatical question factor. And the syntax is then one reflex of it. So then, then from that perspective, the question becomes, in the case that what you are showing actually is quite the opposite to the left over pre-purposing, but rather, the original function. But actually, that's our access to what is underlined is not the language, which then are no longer languages, but reflex of something else. But then the question becomes, what is something else? Yeah, I mean, that is very plausible that really, you know, always you tend to think of it as well. What's in the Philippines is the older thing and everything else has changed from that, but it's especially given how isolated Engana is, it's possible that it preserves some kind of earlier pre-symmetrical voice type system. It's totally possible. I don't have a good answer for what it is, if it's not that, but yeah, something to do with tense aspect, maybe, maybe I don't know. That's fine. That's fine. That's fine. Yeah. Just related to historical change. I don't know what I'm missing it on, but this seems like a lot of change for 100 year period. I don't know if it feels the same way. I'm wondering like, so does that have to do with, you know, what is there a more complex dialogue situation or levels of language or were there pre-Austronesian groups there, which of course would correlate with the lexical change between totally different and things other things that were going on that have influenced this huge amount of change over time. Yeah. It's really a lot of change because it's probably two or three generations, right from the 1930s to today. And, you know, the words look really different, largely because you've lost the final vowel. Once you factor that in, it's not, but then you get these other changes, like the age construction. I don't know why there would be so much change in this particular context. Is it somehow to do with the fact that there has been a lot of language contact? There are, I don't actually know what the relative proportion of Engano speaking people to non-Engano speaking people in the island is, but I think there are lots of people who don't speak Engano on the island. The vitality is much less than some of the other languages in the area, according to the most recent census at least. So can that have something to do with the fact that they've got the change? Yeah, I don't know. But definitely it looks like a lot of change to me. But why would language, I mean, this is the general question, why would language contact lead to losing the final vowel? I mean, there's always theories about who are, are there other people learning this language? As a second language, I have other influences that bring with them, but that's always the story. Lucy, do you have a comment on the story? Actually, I have a whole page of comments for that. There's two talks in one, there's three of them, thank you. But I think on the vowel one, it reminds me of Hawaiian passives, which is more of an intimate language, so the Hawaiian passives, you didn't have to analyze. There's all these random vowels which makes it particular. But what we have to do is, we have to say, well, the passive is the original one, and there's nothing done about it. And the active, and that's the vowel, it's not really losing the final vowel. So that, you know, so why that question, then the logic, but it is more apparent, because it's funny, having strange vowels, when I've been lost in places, where you don't expect, it looks a little bit like Nostronesian. Very possible, very possible. Okay, then let me, and then let's go back to Chris's question, which I think is interesting. And the question why there's so much documentation, and that's why I think, I think what you're doing is really interesting. It's a different from, maybe more, more well-known scenarios of all texts, because it's such a strange, as a collection context. These are all kind of long and complex, right? These are little German and Dutch missionaries and anthropologists going there, and they come with a very clear expectation of a language of life. And they're all trained in Latin, and they are trained in African music. So maybe what you get in the description, actually, is an idealized version of what the language should look like in the view of the missionaries. So in Southern Africa, there was the famous hospital in Kula, Kula comes from Kuala Lumpur, a great of English. He went back to Southern Africa and found out that all the croissants, because they couldn't speak their croissants, they didn't really understand how good well. So then Kula described his own version of what he thought the croissants should look like. And that he really likes that. It's very sweet that he doesn't mean bad, but of course you can see the mindset behind it. So maybe what we're looking at there in the whole description is some sort of ideal of what ought to be like. It was methodical, because it's based on that. So the change you see is partly artificial, because it never was that way. But of course, what is ultimately so good. So that's why anthropology is so interesting, I think, what we're doing, because it's very unique. There's a lot of documentation called a classical description type situation, which is sort of difficult to choose, and then you could, I guess, I'm very, very impressed with that, but the project could ask, who are these people like Kula, who did these two other things? They're not them, but I think it's a big, humble, Austronesian person. How did he end up there? What was the preconceptions? Why did they go to this place, not the other? So it becomes that very deep description almost, in part by then said, like on very different structures. Yeah, I think that's another way of making the materials of classical, and then kind of background to those projects, and I think that would be very useful. Yeah. Yeah. Just very narrowly on the final vales thing. I mean, final vales, they draw, that's what they do, right? But at the point of time that they draw, they draw a gradual starting in an agro stage, and building on looks at this point, maybe Kira was describing, maybe final vales in the variety he was claiming to document, as it were, or he was very often dropped, but he prefers full force when they're speaking carefully in this possibility. Yeah, yeah. That's pretty likely. I didn't have another question. Just a sort of quite a small round of question. Yeah, forgive me because I forget the contents, but you talked about functional marking, I just wanted to know what exactly the criteria for that. Yeah. In relation to assessing alignment, it's looking specifically at the main or transitive clause. So the idea is that transitive clause, according to various people would have particular semantic properties. So the unmarked one would have those exact properties and anything that varies from that would be marked in some way. Also, in terms of discourse, it would be the most frequent one in discourse that would have particular discourse pattern. So there were some kind of metrics for assessing that, but functional rather than full. So it wouldn't rely on a kind of formal definition of transitivity, but rather semantic discourse definition. That hasn't really answered the question in detail, but yeah. I guess I've just become slightly allergic to the unmarked. Yeah, I think, yeah. Yeah. Yeah. You could, I don't know. Do you like prototypical? I mean, people are really, yeah. Yeah, I have also encountered that people are not, market-ness wasn't very fashionable, but it seems like a nice tool. Yeah. I mean, I think the point is, I think the trouble is, often people use it to say something very vague but possibly ultimately meaning, listen clearly, that's not the case with you, but in the other cases, like in your case, it breaks down to some of the precise criteria. Yeah. Maybe the criteria are the wrong criteria, but there were a set of kind of criteria. And you could see a difference. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. I mean, I guess it was to use it. Yeah. Just a way of choosing between one or other of these constructions really for the purpose of deciding what your alignment should be. Yeah. Double check. There's no questions here. See any questions, anyone on mine? Are there any final questions from anybody else here? Just kind of building on what we said before regarding the historical documentation. Yeah. Thinking about what you said about missionaries in there, like previously notions of how language should be. I know something similar happened with a lot of Native American languages and they were documented in a way that actually, I think changed the way they worked like down the line. I just wondered if maybe you said it's like a lot of change in a few generations in this language. If maybe the historical documentation could have triggered some of those changes? I don't think that speakers would have access to these materials because they would be in Dutch or in German. So I don't know that they could have had to kind of do some sort of effect that you're describing. These people document and then maybe we're traveling and maybe carrying some of these notions with them that they previously documented two parts where the language was slightly different might have influenced something. Maybe. My sense is that a lot of the changes are making modern and more similar to Indonesian and that therefore it's likely a kind of contact effect of Indonesian being the national language and also local varieties of Malay and the local languages that are important languages of regional communication and all have these Indonesian type structures. So my sense is that a lot of the changes that you see in contemporary Angano are probably tied to contact in that way more than an effect of the early documentation. Could it be possible that these historical documentors were not that accustomed or not that used to Angano. So maybe some of the sounds that they were documenting they couldn't distinguish between several finds that they maybe sort of had. That is really tricky when we're looking at the historical resources particularly when there's a difference between so recently we've been we've mainly worked with the Kaila materials and recently we've been trying to kind of incorporate the earlier health information materials too. And it is really challenging to work out what is just a very and just you know we're not represent Kaila represents a glottal stop with an apostrophe we assume everywhere he heard it but that's an assumption. And health sometimes represents a glottal stop with a Q and sometimes that will well in cases where Kaila has one he doesn't have one. Is that because he assumed if he wrote two different vowels next to each other that there would be a glottal stop added. And so there's no need to write it. Or was it not there are there two different strategies for having the vowel sequence and dealing with that. Yeah, in terms of so yes I think it's entirely plausible that the kind of native phonologies of the people documenting the language affected the way that they transcribed the test. I don't know the Kaila documentation seems very systematic to me. And almost always when we have the contemporary form that corresponds to the old form it corresponds exactly and I don't know whether that is because I'm using the historical I'm so familiar with the historical materials that I then hear it in the way that I think it ought to be pronounced based on that that is possible. Or whether it was just you know pretty systematic documentation. Yeah. If we could, there are we we know that there are manuscripts of the Kaila material and various people have because the when the dictionary was published as mention of having seen these manuscripts but we don't know what happened to the manuscript so kind of the major aim of our life for the last however long it's been to track these manuscripts down and I hope that it might tell us whether that will be possible. Yeah, it's a good question and I don't have it. It's very hard to tell. Maybe we'll wrap it up there. Now anyone who wants to continue to discuss things is free to join us afterwards as we'll just take the discussion elsewhere but let's say thank you to Charlotte very much.