 So what I decided to do for this audience is to focus on one of the many controversial points in my recent reconstruction of ProtoBask, or the oldest form of the Euskadian language. Sorry, I know as I was here. Just for those of you who don't do much with Bask, some of the reasons why it's thought to be sort of a difficult language to work on are no longer probably that relevant. I mean, the language is still spoken by very many people, under a million, but still it's quite healthy. It's spoken, as you probably know, in the Bask country. There are at least seven major dialects. Some of the big dialects have disappeared in the last 100 or so years, but there's still a lot of diversity. There's also the Aquitainian language, which most people agree is a relation of modern Bask. And this language is attested in Latin text, dating from about 0 to 300 before the Common Era. There are also many place names from antiquity that can be used in the reconstruction. I think some of the medieval data will be discussed in this workshop as well. And yes, well, the language is considered an isolate by many, including those who are producing the etymological dictionary that will appear this year. There's been a lot of work on Bask historical phonology, the best known being the work by Martinet and Michelina. There's been work on Aquitainian. There's been work on the root structure of proto-Bask, on the accentual system, starting with the different dialects and reconstructing back from there. And there's also been work on the phonetic basis of sound change in the history of the Bask language. And again, you'll hear more about that later in the workshop. And the reason I said that I think the situation has changed from early in the 20th century is because there are now a great deal of lexical resources on the language that are searchable, including the 16-volume orotarico escaricegia. Asque is 1905, 1906 dictionary, which is full of useful cultural information, is available electronically. And now we also have collections of Aquitainian data. Actually, the last thing on this slide is the collected digital version of the cartularies compiled around 1100, which contain place names and personal names, many of which maintain sounds that have been lost in modern Bask. So for the past five or so years, I've been working on reconstructing proto-Bask, which would be meaning methodologically using both methods of internal reconstruction and the comparative method where the Aquitainian and dialect data is available. And much of this builds on the work of Michelina, in particular, as you'll see, the debu-calization of initial P-T-K to H plays a central role in my reconstruction. It's something that Michelina thought was going on, but I think I've found many more examples of that. So what I'm presenting today is a very small piece of the big puzzle relating to the status of sibilance in the oldest form of the Bask language. So I'm just reviewing here some of the problems with reconstructing isolates. I think most of us are aware of these here, so I'm just going to go quickly the comparative method. I didn't know whether we would all be Neogramarians using the same method, so these are reviewed here. But there is a problem in working on Bask, and that is in identifying inherited words, because we know from the known history of Europe that Euskadian has been in contact with at least three branches of Indo-European. The Celtic languages, the Romance languages, beginning in about 200 BC and continuing, and then Germanic languages as well. So a big part of the difficult work is separating out potential loans from inherited vocabulary. And of course, it really helps to have a good sense of what the proto-language looked like to start weeding these things out. But I think you have to try to be doing both things at the same time. In the etymologies that I offer at the end of the book that I've published, I've tried to stay away from any term which has what looks like a potential cognate in Celtic, Romance, or Germanic, unless I have a good argument why it could not be borrowed. So here, words that appear to be native, maybe loans, and seeming loans may be native. So here we have a word that you might see later today or tomorrow, Ibaii. That's medieval Basque for river, so it's attested with that H in the ultimate syllable. And this is considered by most Basqueologists to be a native term. But there is a Celtic form or proto-Celtic form which looks very similar. We can't definitively rule out that it could have been loan from some Celtic language. We have other words. If you go skip down to the bottom here, like the word bara, it looks like the word for bar or barrier in vulgar Latin. But I think there are reasons to believe that actually bar is a native Basque root. It occurs in constructions like this term elbar, which has a semantic range that's sort of been extended outside of barring with an obstacle. But what's important about the word elbar is the initial part, the el, which I believe comes from del, which is a proto-Basque root. So this just shows you the difficulties inherent in this kind of exercise, especially if it turns out that proto-Basque is related to proto into European. That means there are going to be roots that look similar in the two languages. So as a background, and some of this is on the handout, the proto-Basque vowel system is a five-vowel system. On this slide, the diphthongs are in parentheses because, in many cases, things that look like diphthongs in modern Basque would have had an H or an N between the two vowels. But I'm following Michelina for the most part in this five-vowel system. And the big differences that come up are in the consonant system. So here you see Michelina's 1977 system. There's no P, but there's TK, BDG, and then at least two series of affricates and fricatives. That's the focus here. The two fricative sounds. You see that these are also reconstructed in Nakara's work. And they also are in the forthcoming Basque etymological dictionary. There is a contrast in all of those systems between an apical S-like sound and a laminal S-like sound, which still persists in some modern Basque dialects. And that is reconstructed back to the oldest form of the language. I am contesting that. This is my revised consonant inventory on this slide. As you see, there's only a single sibilant. So the problem here is, how do we get from a one-sibilant system to a two-sibilant system? And that's what I'd like to talk about. Summarizing agreement with earlier proposals, everybody, I think, agrees that Proto-Basque had a voice series BDG, sonarance NLR, and H, which was not reconstructed by, for instance, Trask. He did not have an H. That there was a single rhodic in Basque, not two is in the modern language. And that all consonants except for this rhodic could occur in the onset. Codas were limited to R, L, N, S. And in the coda, you could have R, L, N, followed by S. The differences here between my proposal and others are I'm arguing that Proto-Basque had an M, that Proto-Basque had a P, a voiceless bilabial stop, that Proto-Basque stops were aspirated. I think you'll hear more about this when Kanda Egretsegi speaks. And the focus today, that there was only a single sibilant in Proto-Basque, and that the origin of the contrast between the apical and the laminal sibilant originates from sibilant consonant clusters in the Proto-Language. So in reconstructing Proto-Basque, I have come up with sort of syntagmatic versus paradigmatic properties that aid reconstruction. Syntagmatically, most of these are the same as those used by earlier linguists such as Michelina, that you could have a bare root, a root in a compound as first or second element and reduplicated roots. What's different are the paradigmatic properties. I have identified what I think are two well-attested prefixes. One which is a nominalizer, ha, often expressed as ah in modern Basque without the H. A collective prefix, H-I, the verbal prefix H-E, which usually again is surfacing without the H, is a verbalizer which has been well recognized in Basque historical linguistics as has the participial agitival suffix. And then the two elements at the bottom, number four and number five on this slide, which are new, is an S which serves as a nominalizer, suffix and an S prefix which can produce an SC cluster from a single consonant initial root. And here this is illustrated with the protobasque root bil, which I think is going to appear in the Basque etymological dictionary. It's a well-recognized root meaning turned around as in the modern verb ebili to go or go around. And you can see that this root occurs with the ah prefix, the A prefix. It occurs as an agitival or participial form. It occurs in the word, I believe, it occurs in the word beats, which means like to meet or to come together. It can also be to round up. And perhaps even in the word si, siya, which is the word for belly button, okay. Another small difference between this approach and others is what roots look like. So I've identified, yes, monosyllabic roots, but also disyllabic roots. And you see in the first column, the last two entries here are roots that are claimed to begin with SC clusters. Okay, so let us talk about the single-siblin hypothesis. All the previous reconstructions of protobasque that I'm aware of using the comparative method, I should say, pause it, a contrast between an apical S and a laminal S that's continued in common Basque. And the motivation for this contrast are minimal and near-minimal pairs in the modern language where the two sibilants contrast. So we have words like shu for fire versus su. And as I said, in many modern varieties, the contrast is preserved, okay. That between the apical and the laminal. This contrast also exists in medial position, in intervocalic position, and in final position where the sibilant is africated, okay. So this is a general property of Basque that a sibilant fricative becomes an africate in word final position, okay. So from hach, you get hach. And so this is the big stumbling block to reconstructing a single sibilant. These two sounds contrast. Nevertheless, I believe there are five distinct types of arguments for a single sibilant in the proto language. Okay, the first argument here says, well, if you have a vowel, a sonarant, and a sibilant, you never get a contrast between the apical and the laminal in this position, okay. So that's one position where the two sounds do not appear to contrast. Furthermore, there are synchronic alternations between roots or stems that show a sonarant followed by the laminal, z, where there may or may not be a sonarant present suggesting a sound change where erotic was lost in erotic sibilant cluster. I'm gonna go through the evidence for these. Third, there are root doublets where a z in one form corresponds to an sc cluster in another. Fourth, there seems to be a distributional asymmetry between the s and the z sound. The z sound seems to be at least twice as common in basic vocabulary items. And finally, there appear to be related stop and laminal sibilant initial roots where, well, that capital T represents a historical PTK that's undergone in most cases, initial debucalization. Okay, I mentioned the initial debucalization at the start as something that Michellena had discussed briefly in his work on bass historical phonology. And so I spent a lot of time trying to dig up roots that when they were not in initial position had initial PTK in initial position. They have H to support this general sound change of PTK to H in initial position. If we see then T or PTK with z initial roots, it suggests a possible spustusca to z, the laminal sound change. Okay, so the first thing I mentioned were inherited monosyllables with vowel sonar and sibilant rhymes where there's no contrast between s and z in this position. Here are some roots that I have reconstructed. You've already seen the second one, that root, pil. Okay, and what you notice here is that after a sonarant, you find the laminal africate, not the apoco, the t-z form. So the suggestion here is that at the bottom, these all derive from sonarant plus, well, but what that really is is s, right? Because the africation is a late rule. So to go through the steps, right? The distributional asymmetry suggests a single sibilant, s in the coda. There was a sound change then of coda cluster s-laminalization, taking s to a laminal when it's preceded by a consonant in the coda. And then the common bass final africation, which results in now a contrast between two africates as opposed to the two fricates. Just to show how this works, right? We have a form like protobascache with s not in a cluster versus the word for alder, which has an s that is in a cluster, okay? And what we get on the surface is two africates that are distinct because one originates in a cluster and one does not. So then the obvious question one poses is, wait a second, you said that the africates contrast post-vocalically. So what do we say about words like guts, right? This is the word for salt and huts or ats, the word for finger. Well, in this case, I'm arguing that there was a sonorant there. It's been lost, but we have other arguments that it actually exists. In the case of the word for salt, gud is a root, which means grain, okay, as in a grain of salt. And there's a lot of evidence supporting that root in bass. So I said at the start that there was other evidence for this coda cluster simplification, okay, namely actual stem variants within the language, okay? So you can see that whatever the simplification is, it's sort of continued to a point where we can recognize it. So we have the word, for instance, ats, bear, these are modern bass terms. The word for badger, which seems to be derived from bear, right, has many forms actually, but one of them of interest to us has an R and one does not have an R. And the one that does not have an R is showing this laminal fricative exactly where we would expect it, right? Because historically the claim is it's laminal because of the preceding sonorant, okay? The same is true for these forms, for aldergrove and the word for butter, okay? So these alternations motivate a prosodically conditioned coda cluster simplification. Notice the simplification is going on in diselabic or longer forms, okay? So the actual exceptions, I would say, are words like gotts and ats. We don't expect the R to be lost in the monosyllable, but it is. In diselables it seems like it was a fairly regular process. Okay, so again, just showing how this would work. In the case of the finger paw word in the last column of this little table, you see the reconstruction with a root pa, which means like foot, actually. And then a stem, final R, the nominalizer S, okay? So from this we get things like modern pats atts or a patch in the form with the historical ha prefix, okay? The other two words you already saw. With the coda cluster S-laminalization process and coda cluster simplification, we're now in a situation where actually we can show there's no evidence for an S versus Z contrast in the coda in protobasque, right? All of the instances of the laminal can be derived from clusters, laminalization in clusters, okay? So we're now left with the question of, well, what is going on in other positions of the word, okay? The, in medial position, my claim is that the medial position of these roots are cases which were historically final. They've had suffixes added. So the position we're left to deal with is initial position. The only other position where S and Z could contrast is in the syllable onset. S is well attested in the onset, sometimes affricated to tz, and you have some examples here of what I think are true continued S from protoforms with S as in the word for fire or the word for moth, see? But obviously you see there were words that begin with Z in the modern language, so what do we say about these? Well, the third argument for a single sibilant is that there are things that look suspiciously like doublets in the language, okay? The pair in A on this slide, you have two words, ashtun, which is the common word for heavy. This would be a word that would be on a Swadesh list, okay? It has an S-T cluster exactly in the position where I would expect a root initial segment or series of segments because A looks like the ha prefix that I mentioned. And then you also see there is a word, not very common in modern times, a sun, which means full loaded pregnant. So there's an overlapping meaning here between the pregnancy of people and animals. There's a semantic field, which is not uncommon cross-linguistically, right? Heavy meaning, pregnant, weighted down, and so on. It very much looks like a set of doublets. And the same is true for the word for skin shell bark, a sun, okay? I'm comparing this to that Ska cluster in Asaska, assuming that there's been sibilant harmony in this word. Sibilant harmony is a common process in Bask. Okay, there are more of these doublets in the languages, in the language. Here's just a few more sets. Istil is mud-filled puddle pond effluent sindu, means to dirty or muddy. So the idea here is that S-T-I-L is the historical form of this root. Okay, I spent some time talking about these things. That there are doublets, so if we have doublets, we propose then Sphastuska clusters. What happens to them? S becomes Z in a cluster, and then we have simplification of the cluster where the output is just the laminal sibilant. On the handout that you have, there's a paragraph or so about the phonetic basis of this kind of sibilant retraction. Cross-linguistically, there's been quite a bit of work on the phonetics of sibilant retraction in clusters. It is not an uncommon process. I know we have many people here working on Indo-European languages, so I'm sure you can think of many cases in Indo-European, including things like German, Ska, Tusha, okay? That's exactly the kind of sound change I'm talking about here, except that in Basque, all three clusters, Stospa, Ska, became something like S. I mentioned that there are asymmetries in the distribution of the sounds S and Z in Basque, and it's hard to express that by just showing lists of words, but there are many words like za, sa bada, zain, and so on and so forth, which would be considered inherited Basque words. They begin with Z, and I imagine this will be a big letter in the Basque etymological dictionary, because these are all considered native words. There are many more beginning with Z than beginning with S. And finally, and this is sort of very interesting property, there are roots that appear with Z, which then can be related to things that occur on the surface with a P-T-K that has been debutalized in initial position, so it shows up as H. And I think, yes, I've included here sort of my favorite word because if this etymology is correct, it's very interesting also from a comparative perspective. So I think the word for 10 in Basque, which is at the top of this list here, a la, has puzzled people, there are many different etymologies of the word for 10. Basque has a base 20 counting system, so 10 is sort of only halfway, but under my analysis, this word for 10 actually means at the top, or top of the count. The question is, on what basis do I reconstruct a T for this form? Well, I do it on the basis of some other words that don't really look necessarily like they're related to the word 10, but the one that really caught my interest was the fourth entry down here where you see tamar in a word, shash tamar, which means residue remains, waste, flotsam, the things that float on top of the water, the dirty things. Shash in Basque means dirty, and it shows up in lots of words, it shows up in the word for broom to clean up and so on and so forth. But the point here is, literally this would be, mean something like dirt at the top, okay? So it's an attestation of tamar, which I'm relating to amar, the word for 10, meaning top. There are other words like gardama, which means the, seems to mean the milk fat, and it means milk fat in modern Basque, but literally the drops things at the top, and what's interesting here is that there's a very common Basque word, samarra, which is the Basque sheepskin coat. It has many other meanings in Basque, including skull cap, frontal piece put on an ox, a blacksmith's apron, a smock. It is a piece of clothing or paraphernalia that goes on top of something, okay? That seems to be the general meaning of the term if we put everything together. So why does it have a z? It has a z, I'm saying, because it is historically derived from an ST cluster. So this is the point at which there's some other forms like this that I've given you information on. This is the point at which I say, well, this is all very interesting, the internal reconstruction of ST clusters in Basque because with things like spustasca, we have more material for a comparison between this language and this proto-Basque language and proto-Indo-European. And for this, because I'm out of time, I can refer you to the handout and to more information in the 2018 book. And thank you all very much for listening.