WEBVTT 00:00:00.080 --> 00:00:05.660 Hello. In the Crown of Aragon 639 years ago, 00:00:06.280 --> 00:00:13.439 it was March 1, 1387. My name is Jonathan Seyfried. 00:00:13.679 --> 00:00:16.859 I'm a PhD candidate in history at the University 00:00:16.859 --> 00:00:20.899 of New Mexico. And this is the Historian's Notebook, 00:00:21.359 --> 00:00:25.640 a podcast about how history gets made. Season 00:00:25.640 --> 00:00:29.719 1 is titled Molt Cara Companyona. We're looking 00:00:29.719 --> 00:00:32.659 at a document from each day of the first year 00:00:32.659 --> 00:00:37.259 of the reign of King Joan I of Aragon and Queen 00:00:37.259 --> 00:00:41.020 Violant de Bar. Let's begin today's document. 00:01:04.750 --> 00:01:10.829 Well first of all we are now in March and No 00:01:10.829 --> 00:01:16.129 more of those lovely double F's for February 00:01:16.129 --> 00:01:23.189 I'm gonna miss those fun double F's but luckily 00:01:23.189 --> 00:01:29.790 for us in the final sign off to Today's document 00:01:29.790 --> 00:01:34.519 there is a double F but it is actually part of 00:01:34.519 --> 00:01:38.019 this. Oh gosh, I don't even know what you call 00:01:38.019 --> 00:01:39.900 them. They're these little phrases. I'm trying 00:01:39.900 --> 00:01:42.340 to get to the bottom of this and actually the 00:01:42.340 --> 00:01:44.239 Latin group that I mentioned yesterday is going 00:01:44.239 --> 00:01:48.099 to help out with this. There's these series of 00:01:48.099 --> 00:01:52.870 little phrases. that occur at the end of the 00:01:52.870 --> 00:01:56.349 documents and it's sometimes there's a name and 00:01:56.349 --> 00:01:59.549 sometimes it's not a name and they're usually 00:01:59.549 --> 00:02:03.750 like one two or three words long and they've 00:02:03.750 --> 00:02:07.090 been a little bit of a mystery to me sometimes 00:02:07.090 --> 00:02:10.330 i can figure out okay it's just one person's 00:02:10.330 --> 00:02:14.629 name but it how does it relate to the person 00:02:14.629 --> 00:02:17.969 who is signed off on the probatum at the bottom 00:02:18.520 --> 00:02:22.039 as saying, you know, I'm authorizing this document. 00:02:22.520 --> 00:02:28.539 Is it somehow related to the format of the actual 00:02:28.539 --> 00:02:31.960 letter that got sent out, the original copy? 00:02:32.460 --> 00:02:34.960 So I'm trying to get to the bottom of that, but 00:02:34.960 --> 00:02:38.680 hey, if you're feeling not quite ready to let 00:02:38.680 --> 00:02:43.759 go of the double F's of February, well, hopefully 00:02:43.759 --> 00:02:48.819 the very last couple words of today's document 00:02:48.819 --> 00:02:55.240 will help you to gradually release yourself from 00:02:55.240 --> 00:02:59.699 the joy of seeing the double F of February because 00:02:59.699 --> 00:03:06.500 that's it no more February 1387 all right well 00:03:06.500 --> 00:03:09.719 that was a little somber but what I can tell 00:03:09.719 --> 00:03:12.919 you about today's document is that it's kind 00:03:12.919 --> 00:03:18.090 of heartwarming And this document is basically 00:03:18.090 --> 00:03:22.830 authorizing an old age pension. So there was 00:03:22.830 --> 00:03:29.810 this guy named Domenico Lull and the header above 00:03:29.810 --> 00:03:34.069 the letter or document is really more of a document, 00:03:34.289 --> 00:03:40.289 not a letter. The heading has the name Domenico 00:03:40.289 --> 00:03:46.289 Lull with the P and the curl around to the left 00:03:46.289 --> 00:03:54.169 for pro. So that heading reads for Domenico Lull. 00:03:54.469 --> 00:03:58.370 And as you get into this document, you realize 00:03:58.370 --> 00:04:05.229 that Domenico Lull is a bailiff or I should say. 00:04:05.389 --> 00:04:11.949 was a bailiff in Valencia who was a loyal and 00:04:11.949 --> 00:04:18.050 trusted bailiff for the state in the city of 00:04:18.050 --> 00:04:22.850 Valencia for many years and has now reached old 00:04:22.850 --> 00:04:29.949 age. And this document authorizes a pension for 00:04:29.949 --> 00:04:34.639 him so that he can live out the rest of his years 00:04:34.639 --> 00:04:40.519 after his loyal service in comfort and have some 00:04:40.519 --> 00:04:44.920 kind of pension to help him meet his expenses. 00:04:45.620 --> 00:04:52.439 So this is a social service document. It's super 00:04:52.439 --> 00:04:56.920 interesting to see. It's not representative. 00:04:58.660 --> 00:05:01.480 There were not old age pensions for everybody. 00:05:02.100 --> 00:05:06.160 You really had to have a high up government job 00:05:06.160 --> 00:05:09.339 in order to have an old age pension. But it's 00:05:09.339 --> 00:05:13.379 kind of neat to see this old age pension sort 00:05:13.379 --> 00:05:18.199 of a social safety net, at least for somebody 00:05:18.199 --> 00:05:24.079 in the middle ages. And just to have this evidence 00:05:24.079 --> 00:05:29.500 of a kind of kindness, a kind of reward for service 00:05:29.500 --> 00:05:35.720 that comes from years of loyalty, I just think 00:05:35.720 --> 00:05:41.180 that it's a nice sweet moment in the documentation. 00:05:42.339 --> 00:05:46.519 Now, I'm romanticizing it when I go into it like 00:05:46.519 --> 00:05:50.269 that. It's totally possible that Domenico Lull 00:05:50.269 --> 00:05:53.189 was a terrible person and was a total jerk and 00:05:53.189 --> 00:05:55.389 exploited people. And maybe one of the reasons 00:05:55.389 --> 00:05:58.750 why he was such a good bailiff was because he 00:05:58.750 --> 00:06:02.310 was using muscle in order to extort people to 00:06:02.310 --> 00:06:07.310 pay up. So we can't assume that just because 00:06:07.310 --> 00:06:09.790 somebody is getting an old -age pension that 00:06:09.790 --> 00:06:13.569 they were, you know, loyal, trusty, and all of 00:06:13.569 --> 00:06:17.670 these positive qualities. But it's possible, 00:06:17.709 --> 00:06:22.689 right? And so that's just the kind of check that 00:06:22.689 --> 00:06:25.370 I need to put on myself whenever I get a little 00:06:25.370 --> 00:06:31.870 bit teary -eyed in a document. I'm actually teary 00:06:31.870 --> 00:06:34.629 -eyed because it's like full -on allergy season 00:06:34.629 --> 00:06:39.149 in Albuquerque right now, so it's not exactly 00:06:39.149 --> 00:06:44.050 connected to to Domenico Lull and his old age. 00:06:44.610 --> 00:06:48.329 Let's talk about how old age is referred to. 00:06:48.329 --> 00:06:51.209 Let's do a bit of a close reading of this document. 00:06:51.689 --> 00:06:56.930 So in line 10, about three words in is the word 00:06:56.930 --> 00:07:03.509 senio. senio and then the word after that is 00:07:03.509 --> 00:07:14.170 pregravitatus so senio is in the ablative it 00:07:14.170 --> 00:07:18.230 could also be in the dative i think it's it's 00:07:18.230 --> 00:07:22.149 clear here that it's in the ablative and i think 00:07:22.149 --> 00:07:25.769 oh i didn't look it up ahead of time i think 00:07:25.769 --> 00:07:34.620 pregravi tus is also oh gosh I'm so yeah trying 00:07:34.620 --> 00:07:37.360 to remember these Latin case endings because 00:07:37.360 --> 00:07:43.079 there's there's different forms of nouns different 00:07:43.079 --> 00:07:45.980 categories of nouns called declensions and you 00:07:45.980 --> 00:07:51.920 gotta know which is which all right so but I 00:07:51.920 --> 00:07:55.120 think actually this is more of like a verb oh 00:07:55.120 --> 00:08:02.759 my gosh This phrase means weighed down by age. 00:08:03.000 --> 00:08:08.079 Yeah, so pre -gra -vi -tas. Oh, God. Pre -gra 00:08:08.079 --> 00:08:13.939 -va -tus is some kind of verb form. And the phrase 00:08:13.939 --> 00:08:18.560 basically is a elegant metaphorical way of saying 00:08:18.560 --> 00:08:24.879 you're really too old to do the work anymore. 00:08:25.290 --> 00:08:29.629 it's kind of a nicer way of saying enfeebled 00:08:29.629 --> 00:08:39.889 and so senio here is the ablative for agedness 00:08:39.889 --> 00:08:46.590 you can say and then pre gravatus is a verb form 00:08:46.590 --> 00:08:51.350 for saying weighed down so you can you can sort 00:08:51.350 --> 00:08:56.190 you can start to kind of see the English descendant 00:08:56.190 --> 00:09:00.190 from these Latin words, right? Because in Pre 00:09:00.190 --> 00:09:06.570 Gravatus we have the English word gravity for 00:09:06.570 --> 00:09:10.350 something about weight. And then, of course, 00:09:10.769 --> 00:09:16.110 in Senio, if you just put an R in English at 00:09:16.110 --> 00:09:19.610 the end of those letters, you have the word senior. 00:09:20.429 --> 00:09:24.470 So that was a really great opportunity to draw 00:09:24.470 --> 00:09:31.570 out some kind of Latin to English lineage there, 00:09:31.570 --> 00:09:38.370 and also to pull out a little bit of an idiomatic 00:09:38.370 --> 00:09:42.210 expression that really clarifies that all of 00:09:42.210 --> 00:09:48.259 this is about an old age pension. I did not pinpoint 00:09:48.259 --> 00:09:51.840 this on my own. Like, if you're having this idea 00:09:51.840 --> 00:09:55.139 that I suddenly got amazing at Latin and just 00:09:55.139 --> 00:09:57.720 read through this and then pulled out that phrase 00:09:57.720 --> 00:10:01.059 as the important one, that would be inaccurate 00:10:01.059 --> 00:10:04.580 because my Latin is nowhere near good enough 00:10:04.580 --> 00:10:07.879 just to kind of skim through that all this dense 00:10:07.879 --> 00:10:12.889 writing and then flag that right away. In fact, 00:10:13.350 --> 00:10:16.269 senio in my mind wasn't immediately connected 00:10:16.269 --> 00:10:19.409 to the word for old man in Latin, which is senex, 00:10:19.750 --> 00:10:21.970 which is one of the first words you learn if 00:10:21.970 --> 00:10:24.809 you're studying Latin. It's one of your initial 00:10:24.809 --> 00:10:28.750 set of vocab that you get introduced to if you're 00:10:28.750 --> 00:10:33.129 starting out with Latin, senex. So I didn't make 00:10:33.129 --> 00:10:37.669 that connection and so it guess who it was? It 00:10:37.669 --> 00:10:44.629 was AI. Yeah. So I was remarking to my dissertation 00:10:44.629 --> 00:10:49.710 advisor, Dr. Gibbs, the other day that it's like 00:10:49.710 --> 00:10:54.590 this podcast is somehow turning into like a journal 00:10:54.590 --> 00:10:59.700 of of the usefulness of AI as it develops and 00:10:59.700 --> 00:11:02.399 as new versions come out and as people learn 00:11:02.399 --> 00:11:05.059 how to use it. And it's almost like there's a 00:11:05.059 --> 00:11:09.059 shadow podcast about the AI. But really, the 00:11:09.059 --> 00:11:11.559 whole idea of the Historian's Notebook is to 00:11:11.559 --> 00:11:14.879 think about the different tools historians have 00:11:14.879 --> 00:11:19.620 when they initially encounter documents as they're 00:11:19.620 --> 00:11:22.919 trying to figure out their argument about what 00:11:22.919 --> 00:11:26.840 happened in the past and also to put together 00:11:26.840 --> 00:11:31.100 in their mind some kind of narrative. So there's 00:11:31.100 --> 00:11:33.980 all sorts of tools that historians have used 00:11:33.980 --> 00:11:40.879 and these new tools with AI are ones that are 00:11:40.879 --> 00:11:44.370 really shaking things up. in the way that we 00:11:44.370 --> 00:11:48.110 can see demonstrated in today's episode. And 00:11:48.110 --> 00:11:50.710 here's Shadow to say hi. Come on up, Shadow. 00:11:51.690 --> 00:11:57.110 All right. So what I did for this one is as I've 00:11:57.110 --> 00:11:59.529 talked about how I've been frustrated sometimes 00:11:59.529 --> 00:12:03.470 by Gemini, because occasionally I know that's 00:12:03.470 --> 00:12:07.309 my frustration Shadow's frustrated too. Sometimes 00:12:07.309 --> 00:12:10.470 Gemini comes out with a wonderful transcription 00:12:10.470 --> 00:12:15.610 of this 14th century gothic secretarial hand, 00:12:15.889 --> 00:12:19.210 and other times it doesn't. So what I decided 00:12:19.210 --> 00:12:22.070 to do as a process that I'm going to stick with 00:12:22.070 --> 00:12:24.889 for a little while, a couple weeks or something, 00:12:25.049 --> 00:12:29.389 is I'm actually using more than one AI in order 00:12:29.389 --> 00:12:33.730 to help out. So I'm starting with Gemini and 00:12:33.730 --> 00:12:37.289 getting a transcription. Then I'm moving to Claude 00:12:37.289 --> 00:12:39.830 and having Claude do an original transcription, 00:12:39.889 --> 00:12:43.460 which is never as good as Gemini. But then Claude 00:12:43.460 --> 00:12:47.200 seems to be better. I know Claude seems to be 00:12:47.200 --> 00:12:52.059 better. Would you please? OK, I pause the recording 00:12:52.059 --> 00:12:55.200 so that Shadow could finish telling me what she 00:12:55.200 --> 00:12:59.659 needed to tell me. And I think she's happy now 00:12:59.659 --> 00:13:02.039 having communicated what she needed to. She's 00:13:02.039 --> 00:13:05.580 still here, so we'll see what happens. Alright, 00:13:05.620 --> 00:13:09.320 what was I saying? Claude is better at thinking 00:13:09.320 --> 00:13:13.320 through. the corrections to the transcription. 00:13:15.360 --> 00:13:20.240 So Gemini is better at the initial figuring out 00:13:20.240 --> 00:13:25.240 of the alphabetical characters written in the 00:13:25.240 --> 00:13:28.539 gothic secretarial hand. Gemini is better at 00:13:28.539 --> 00:13:33.340 expanding the abbreviations. Claude is better 00:13:33.340 --> 00:13:37.179 at catching some things that don't quite make 00:13:37.179 --> 00:13:40.620 sense. And the process for today's document is 00:13:40.620 --> 00:13:44.659 actually a great example of that. So I had Claude 00:13:44.659 --> 00:13:46.940 do an initial transcription of this document 00:13:46.940 --> 00:13:51.460 based on the image that I uploaded. It was not 00:13:51.460 --> 00:13:55.879 good. And then I put in Gemini's transcription 00:13:55.879 --> 00:13:59.460 and said to Claude, hey, here is a transcription 00:13:59.460 --> 00:14:05.279 from another source and please compare it to 00:14:05.279 --> 00:14:09.139 what you did, and to the original image, and 00:14:09.139 --> 00:14:13.879 then reconcile these as best you can. That is 00:14:13.879 --> 00:14:18.720 a process that Claude handled very well. And 00:14:18.720 --> 00:14:22.740 so from there, there were some important corrections. 00:14:23.320 --> 00:14:26.879 for some reason and this is really this this 00:14:26.879 --> 00:14:29.340 i think is super interesting when we start to 00:14:29.340 --> 00:14:32.279 think about how large language models work and 00:14:32.279 --> 00:14:35.320 how they're all operating from recognizing patterns 00:14:35.320 --> 00:14:39.600 for some reason gemini actually inserted the 00:14:39.600 --> 00:14:45.200 word militi the word for knight or soldier into 00:14:45.200 --> 00:14:51.289 the document it's so weird like there's no It's 00:14:51.289 --> 00:14:54.990 not in there at all, but in the first few lines 00:14:54.990 --> 00:14:59.590 it in Gemini inserted the word Mili -ti in front 00:14:59.590 --> 00:15:04.610 of the name of Domenico Lull because it recognized 00:15:04.610 --> 00:15:08.590 in a lot of this other stuff that if we're talking 00:15:08.590 --> 00:15:12.539 about like an old -age pension. Maybe it even 00:15:12.539 --> 00:15:16.299 kind of went back to my earlier document from 00:15:16.299 --> 00:15:19.919 a couple days ago with Berenger Dolms, who in 00:15:19.919 --> 00:15:25.620 that one there was a milliti, a knight. So yeah, 00:15:25.899 --> 00:15:29.179 Gemini inserted that in there. And it's like, 00:15:29.379 --> 00:15:34.299 why are you violating the definition of the word 00:15:34.299 --> 00:15:38.909 transcription? Transcription in no way can be 00:15:38.909 --> 00:15:42.049 interpreted as insert some words. But I guess 00:15:42.049 --> 00:15:44.370 if you're expanding abbreviations, you're inserting 00:15:44.370 --> 00:15:48.710 some letters, I don't know. But that was a really 00:15:48.710 --> 00:15:52.710 interesting output that I noticed from Gemini. 00:15:54.990 --> 00:15:59.009 So Claude identified that and said, There's no 00:15:59.009 --> 00:16:01.850 word mealy tea in the document, and so I'm going 00:16:01.850 --> 00:16:04.870 to remove that as I reconcile the transcription 00:16:04.870 --> 00:16:07.610 that I did with the transcription that you have 00:16:07.610 --> 00:16:11.429 provided me from another source. And both what 00:16:11.429 --> 00:16:16.970 I will say is that both Gemini and Claude identified. 00:16:18.330 --> 00:16:23.070 This as a document that initiates an old age 00:16:23.070 --> 00:16:27.879 pension for Dominico Lull and both of them. pulled 00:16:27.879 --> 00:16:37.340 out senio pre gravatus as a key phrase that identifies 00:16:37.340 --> 00:16:46.539 Domenico Lull as old and in that category for 00:16:46.539 --> 00:16:50.759 receiving an old age pension. So it was a really 00:16:50.759 --> 00:16:55.919 interesting experiment with the large language 00:16:55.919 --> 00:17:00.500 model AI chatbots, and we'll see what happens 00:17:00.500 --> 00:17:04.839 as I continue to use that process. Thanks for 00:17:04.839 --> 00:17:08.319 listening to this episode of the Historian's 00:17:08.319 --> 00:17:12.880 Notebook, Season 1, Molt Cara Companyona. If you 00:17:12.880 --> 00:17:15.539 are leaving with more questions than you arrived 00:17:15.539 --> 00:17:18.819 with, like who got the old age pensions and who 00:17:18.819 --> 00:17:22.839 didn't? I'm really curious about that. but we 00:17:22.839 --> 00:17:26.119 don't have the answer from this document. If 00:17:26.119 --> 00:17:28.900 you have questions like that, I've done my job. 00:17:29.559 --> 00:17:32.740 Because remember, the motto of the Historian's 00:17:32.740 --> 00:17:37.779 Notebook is dissatisfaction guaranteed. Visit 00:17:37.779 --> 00:17:41.720 the website to see an image of today's document 00:17:41.720 --> 00:17:45.839 and additional show notes. and listen again tomorrow 00:17:45.839 --> 00:17:49.099 to hear about the next day in the first year 00:17:49.099 --> 00:17:53.759 of the reign of King Joan I of Aragon and Queen 00:17:53.759 --> 00:17:57.839 Violant de Bar. In the meantime, take care.