 Good morning. First of all, I'd like to thank Nathan Hill for inviting me here. In this talk I will be offering an overview of the contribution that medieval data makes to the reconstruction of protobasque. And the presentation is going to be divided in four sections. I first offer a brief introduction to the history of basket hemological studies. Then in section 2 I will describe what we understand by medieval data in basket studies and what kind of corpus we have. Section 3 addresses relevant data and discussion of this talk. I will present a list of reconstructed words and morphs and show that recently discovered medieval data actually proved that reconstructed forms were right. And finally, if I have time, I will make an assessment of Blaggin's reconstructive proposals. So when we look at the history of modern basket studies, Michelina's name is the unavoidable name. We can talk about an era before Michelina's works, especially his basketorical phonology. Phonetics, sorry, in 1961. And after them, as you can see in this pre-Michelina era, we have mostly foreign scholars who came from other areas of expertise. The interests and goals of this pre-Michelina etymologies were the following. As you can see, they were looking for genetically related languages. In the case, well, they would compare Basque to Caucasian and all those theories. In the case of Rolves, Romanists like Rolves, they would try to analyse to what extent the original Basque culture was affected by Latin or Roman languages and culture. As you can see, there was no real interest in reconstructing a proto-language. Instead, Michelina and later scholars around him applied themselves to more methodologically standard tasks. Michelina would claim the importance of philological work as a basis for any kind of diachronary construction. In this view, etymologies happened to be an actual outcome of the philological and linguistic work, not an outcome per se. One of the latest results of academic Basque studies is the publication of the historical etymological dictionary of Basque, which is about to be published in a month or two. In this dictionary, following Antoine Meillet's classical motto, reconstruction has to be based on both theoretical accuracy as well as on solid philological grants. We think this dictionary is a good example of the second part of his motto, since we have extensively and systematically browsed medieval documents, which is what we are going to talk today. So after this brief introduction to the history of Basque etymological works, I will go on with the next section. In this section, I will make two main questions. First, what do Basque medieval data look like? I will offer you some examples so you can see what we are dealing with. And then I will offer a list of the kinds of linguistic or philological insights that this medieval data offers to us. So I brought here a couple of pictures of a very important category. So you can see what kind of documents they are. And this is a page of this category which exemplifies clearly the nature of most medieval data. They are written in Latin or Romans and then you have Basque place names or person names. All these are names of Basque names. The corpus I've been using for my dissertation for the dictionary consists of more than 300 medieval categories and collections of diplomatic documents. In these categories, as I have said, we have all the most part place names, but we can also find a couple of sentences, at least the words by a pilgrim and later in the late Middle Ages, very brief excerpts of letters and least words. So what do these place and person names look like? I have gathered a few examples in order to show what kind of linguistic evidence they offer. In this one we have the beautiful field and place name beautiful field, so we have an adjective, noun and phrase. In this one, this one is a similar one, but in this case we have a pospositional phrase with gaave, with a pospositional gaave in which Putsu on the gaave would stand for the bottomless pit or well. And we also have things like a number of phrases, such as beleratsuitoreta, meaning the nine fountains. I also brought some person names. In this case we have uxortua, which stands for your wife Emasteona, which is really analyzed as the good wife. We have things like dongarcia apestegico, with apestegi meaning priest place, and co is the relational posposition, that would mean dongarcia from the priest's house. And then we have some funny and interesting names, such as pedoperis garidario, in which pedoperis would, with the Romans counterpart for Peter, son of Peter. And then we have a nickname, which is garidario, where gari is an object incorporated to an inflected verb, meaning which encodes third person datif. So that would be something like Peter, son of Peter, the one to whom wheat flows, or something like that. We guess it was a rich guy. So even though we have only place names, we also are able to find, I won't say complete sentences, but at least we have inflected verbs and things like that. This is not the only example. So what kind of insight do these medieval data and secondary sources offer? They offer as the early statisticians of some words of the basic lexicon. We also have been able to recover some variants that were not previously attested. They also offer a more accurate picture of the historical distribution of dialectal variants. And finally, and most interestingly, at least for the purpose of this talk, they also offer positive proof of previously reconstructed forms. So as for the earliest attestation, I have organized in the table those words and those words, first attestation, abar, atte, borscht and chivta, branch, door, five and piece of iron. In the vast journal Dictionary, which is the one we are using up to now, we have those first attestation data. For example, in the case of branch, we take the, for example, back at least like 600 years and so on. So that's one of the good things that has looked at medieval data. I have also found previously a non-variance of several words. Like for butsu, you have a variant butsu, which will prove crucial for some reconstruction problems. For garai, we have things like garain, arn garahi. For ibai, I found ibahi and ibain, and there's also ubani. For sustrai root, the sustai and susai, which are all up to now unknown variants of those words. So we also have a more accurate picture of the historical distribution of dialectal variants, as exemplified by the word nine. We have the variant bederatsu, which up to now it was only attested in eastern dialects against the form that has been accepted for standard basque, which is bederatsu. And now we have also the bederatsu variant in a central area, very far in basque terms of its actual modern area in eastern basque. And most importantly, as I said, we have found proof of previously reconstructed forms. So the discussion of these forms will confirm the third section of this presentation. I have gathered examples both from reconstruction of the lexicon on the one side, with a discussion regarding words like ascar, big, strong, ebaki, tukot, butsu, butsu. And on the other side, I have also brought here a couple of examples regarding the reconstruction of the definite declension, more specifically the reconstruction of the locative phrases singular and plural. So let's start with the discussion regarding ascar. There we have what Michelina reconstructed. She reconstructed the word as a derived word from assi to grow with a suffix kor, which has a general meaning of the denote tendency. And this word also has the meaning maple tree, for the identity between the adjective and the tree name, you can just recall Latin robur and robustus. So what do we find in medieval documents? First of all, we have the modern place name ascarraga, which is a modern, quite common place name in the Basque country, as well as a last name, which means maple tree place. And the oldest testation I found for that name is ascorraga. So that perfectly fits what Michelina reconstructed. So Michelina reconstructed ascor and we have medieval ascor. A very important article about the non-infinite verb forms. Trasky identifies a suffix key in several verbs. And this leads him to reconstruct the modern ebaki to cut as eban. We have eban and then we have a valence increase in suffix, which is added to the original root. Again, what do medieval data tell us about this verb? So we have in modern Basque, we have the word for trench, which is a compound name, lubaki, which temologically means the ground that is cut. And we have the medieval attestation in that Romans sentence. We have a Basque word inserted there. Mandaba hacer lebanos para calte del agua, which means he ordered to make trenches for carrying water. So there we have another nice example of how a reconstructed word has actually been found in medieval documents. We have Trask, eban, and we have medieval eban. And this, the case of eban ebaki, is indeed a very good example of how Blevins approach fails in different aspects of the discussion. This approach fails in the basic analysis of the word, and it also fails in the general account of Basque data. She takes for granted that the suffix geeky is part of the root and traces this alleged root back to the preprotobasque in the European. This makes no sense since the addition of the suffix, as we can see, is quite recent. So what's the problem with the reconstruction of Gascon putz? Well, this case and its medieval variant, the case of Basque putz and its medieval variant buitsu, takes us beyond the domain of Basque and Iconic linguistics. In this case, Basque medieval data proved to be crucial for the reconstruction of proto-Roman forms. There is a problem described in Romanistic, as described by Foucher, the French-Romanist Foucher. He asks to himself, how comes that the regular U didn't change? The regular change from U to O didn't take place in Gascon. So Foucher needs a variant in southern Romans, which for him is Gascon, in which there is an E vowel in the first syllable that would prevent that change to happen. So the problem is, as he himself acknowledges, that there's no trace of that vowel in Gascon. And he asks that, he asks that, in this region he didn't have an E vowel in the first syllable, or he had an accent in Latin. So the contribution of Basque medieval data again proves crucial for solving this problem, since there is a variant of Basque word putzu, which fits exactly what Foucher needed for the proto-Gascon form. So the E necessary for the Romanist proto-form is not attested in Gascon, but is attested in Basque. So far these were the examples regarding the reconstruction lexicon. I will now turn to nominal morphology, to two examples of the reconstruction of the definite declension. I have summarised here the discussion of the locatis singular, as you can see according to Jacobson in 1977. A definite phrase such as a chain at home in the house. He would reconstruct a lost consument, which he thought it was a voiced stop. And the problem is that if we accept that A is not an article, but the remnant of the animacy marker, which in Basque is Ga, which doesn't fit very well with what we know about those phrases, since they are definite. So in 2006 I offered an alternative explanation for the lost consument. And based on a very traditional view I simply reconstructed an aspiration, which is the initial consonant of all Basque demonstratives. So this way we simply say that the definite article is A, is also in locative phrases, not the remnant of that animacy marker, but it is simply a demonstrative. So we are simply dealing with the grammaticalisation of a demonstrative phrase into a definite phrase. Now again we have some nice data coming from secondary sources, in which the locative singular phrase shows the lost consument. The model form for a phrase such as Neurean in mine is as you can see Neurean. And we recently found an attestation from secondary source, which actually keeps the original form of the locative definite phrases. So again what we had reconstructed appears in secondary data. So the case now I turn to the case of plural locative phrases, which is kind of similar to the previous one. According to Schubert in phrases such as Echetan in the houses, the Eta in the plural local cases is of Latin origin. So it's the collective suffix Etumeta, which somehow entered the basket, definitely the clenching system. But again following a fairly traditional view of the grammaticalisms of the grammaticalisation of demonstratives into articles, I proposed that in definite phrases such as Echetan in the houses, we simply have a grammaticalised demonstratives, we don't have any kind of foreign affix there. So it's again a case of grammaticalisation of a demonstrative phrase into a definite phrase. And again we find support for this reconstruction in medieval data. So in the place names such as Eroheta, which would mean something like the roots, it's a place name, we would find the original aspiration, which is what we reconstructed. So summarising I have shown a series of examples in which reconstructive work has proven to be correct, thanks to recently discovered medieval data. For Asgard we have the reconstructed form Asgard, which is the one that we actually found in medieval records and so on and so on. So now even though it's not the specific aim of my talk, I'd like to offer a brief typology of the mistakes of our recent reconstruction of Proposal of Protobasque, which is the one that we actually found in medieval data and so on and so on. So now even though it's not the specific aim of my talk, I'd like to offer a brief typology of the mistakes so the author shows a complete lack of philological knowledge. The author takes back to pre-proto in the European Basque words that only exist in the mind of some modern lexicographers and the author also takes back to pre-proto in the European words that are well known neologisms. There are also clear long words from Gascon or Spanish that are treated as if they belong to the inherited lexicon such as Usta, Zango, Mui, which are clear. Usta is from Gascon, Zango is most probably from Spanish and so on. So there are wrong analysis of compound words such as Landelge and Ospe in Delge. The author miscuts the word reconstructing Delge which makes no sense once we know that we are dealing with Landa which is ultimately we think Germanic word and Elge. For Ospe fame, we clearly have a compound norm, a compound noun which clearly divides between a morpheme and lexeme, orchan and pear. We have a recurrent conception of basque morphology such as the previous case I showed you with Evacchi and this case with locative an which is taken altogether as if it were a locative suffix. For me the worst part of this approach is that from a more theoretical point of view there is no attempt to answer any of the questions that have been described in the field for years. So from the point of view of fabasque colleges this is a huge step back in what has been achieved in the field during the century so we come back to the initial discussion in which we were looking for genetically related languages but not actually trying to answer problems that have been described in basque linguistics. And I would like to finish with a methodological reminder of what we are doing here when we are doing historical linguistics which is very nicely put by Thomason who will just read it. To be useful to a historical linguist a hypothesis of genetic relationship must be fruitful, a valid genetic grouping will permit reconstruction and thus lead to a better understanding of the member languages and their histories. A genetic hypothesis does not lead to insights of these kinds which in my opinion is the case therein it is sterile and within linguistics useless. That's all, you have the reference there. That's all, thanks.