0:07 Ok, the point here is I will introduce digital, how it is applied to literary studies, well, of course, it's annoying to digital itself. We need to look at the practice of literary criticism that we usually apply, yes, academic circles or in the context of popular literature, such as what is referred to is a reading model called close reading, yes, we know that all of us from the literary environment have certainly read, yes, those who offer a close reading model, for example, memb, which is the one that is in the context of a TS, for example,
0:54 eek Bui, filled in the danus, yes, the model of the approach can be criticized, yes, by, for example, people like Frank Pomoreetti, a critic, he is actually a comparatist, yes, he is an expert in World literature, so the study of world literature, so comparative literature, and he sees that in close reading there is a kind of hidden assumption, namely the assumption that what is read is something canonical, so something that is considered very, very important and very, very few in number and eh, this can no longer be applied when the corpus is so large, for example, in the context of World literature, in world literature, the corpus can be thousands of millions, so there are so many and it is
1:40 impossible for us to only focus on the best works that are considered So the best Call by for example Harold Blom Western Canon so we have to expand the canon the spirit of democratization of the canon should be All of them are read well but if so the canon or the corpus is expanded then the approach itself becomes inadequate we can no longer use close reading because human life is limited while what must be read in democratic assumptions are millions of books at once Well then a methodological challenge arises How to then do a reading that no longer idolizes the canonical bias eh what is close reading like Well
2:25 then he offers as an intellectual provocation a model called distant reading he himself is a thinker a literary critic literary researcher from Stanford made one called Sebai Stanford literary bers with his colleagues literary laboratory where many scientists there mathematicians were taught discussions to formulate an approach that is more for reading sra leat mesakuk so the way is for the critic no longer to read the literary work directly because there are so many millions of them but he reads the results of machine mapping
3:12 of the literary work well this is an idea that was initiated by moreti 2013 yes and then now there are many developments in the big title of digital humanities the core of digital humanities is a humanities approach that uses digital methods So it's not a humanities study of digital culture so those are two different things yes digital humanities is actually how to use digital technology as a method to enrich humanities research the focus is on the method then the pattern is the approach is macro in nature often see a large evolutionary pattern in a large cultural range such as reading netizen comments on the internet on Twitter for example over the
3:58 last 5 years the number of posts That's definitely there are millions more eh I once took data from Twitter all the posts in Indonesian that contain the word literature in it the number is very large yes can be up to 20 million posts yes so the file that we work on is no longer Microsoft Word but Excel yes eh rows or rows of Excel like and another pattern is that it is cross-disciplinary in nature using a mixed perspective from computers from statistical science and so on and eh even though humor is usually interpretative in nature yes interpreting and not explaining laws like natural sciences but in digital humanities which are
4:44 explanatory in nature, searching for those laws is possible, for example, there is a researcher named manov, he is looking for factors of aesthetic taste. What is it that works behind every Instagram post, he is looking for, eh, a kind of law of netizen taste. G Well, that is also one possibility, there are many tools, the most frequently used, the easiest and free is the voan tool, it can be accessed by everyone, so we will practice that later, okay? Well, the main key to digital humanities compared to conventional ones is that in terms of conventional methods, it is close reading, digital humanities is dist reading, the unit of analysis is also different, if digital tends to focus on one work or detail, human, Eh, that was
5:32 conventional human, now this digital humanities can focus more broadly, like talking about shangre or talking about the period of literary periodization, it is very possible to do with digital research because the number of reads is very large, then the focus is also different, conventional ones look for rigidity, digital ones look for general patterns and tend to be different at the theoretical level, yes, in conventional humanities, theory is very dominant, yes, it drives everything, for example, new history, yes, that means the theoretical framework about, for example, human history, so it's like a mantra that we We hold everything in the conceptual scheme, it's digital, it's not diaend to Mes, we see the pattern from there, we look for the right theory, so it's
6:19 buor and its nature is distanced, there are experts in computer programming, experts in literature and others, so it's impossible for one person to do it, it can be done here, but usually more through a team, the basic inspiration of all digital humor developments in various forms, including those done by proreti, is actually the foundation is in linguistics and here we meet, what is an important meeting, yes, between literature and linguistics, which are usually distinguished, yes, one tends to feelings, one tends to mathematics, language, but in digital Hum, both
7:05 meet. Where do they meet in computational linguistics, especially what was proposed in the 1990s as the distributional hypothesis, namely, eh, one hypothesis at that time that was proposed that actually the meaning can be approached based on the context of the emergence of a word, yes, if a word appears frequently in a certain context of emergence, then its meaning will more or less have similarities, this was stated by Zelic Harris in 1954, yes, words that appear in context Similar ones tend to have similar Magna so to be able to know Magna we have to know the context well interestingly eh this hypothesis it can be operationalized meaning it can be operationalized is can be counted how many times similar context appears
7:51 that similar context then operationalized as co-currence of the same word in the corpus so that the meaning of something qualitative can be approached quantitatively by calculating its proportion that is context Eh co-occurrence earlier was the appearance of the same word so John said a by company it so he we will know a word based on who he hangs out with if he hangs out with what words then his meaning will be close to the words around it well this is a hypothesis that in the years -s could not be done yes Kar he is a theoretical construction But when the development of AI technology and so on
8:36 all these hypotheses can be applied because the technological techniques already exist so simply the meaning is a kind of concurrency of symbols that have no meaning is also close to the basic intuition of structuralists yes So the point is eh language is a system of differences so to be able to know the meaning of a word in language yes we do not have to know exactly what the meaning of each word is we only need to know the distribution of the differences that can then be calculated statistically and operationalized through the machine so when mapped in the process of creating that person Yes we can see that on the far left there is a poet eh Novelist artist in general
9:22 they create various kinds of works in various ways there are words sound Image and so on well all of that can then be processed into the machine through ee mapping What is called tokenization yes so see the smallest element in the corpus what is the word for example yes or letters like that one can be selected like that yes For example word then this word how can the machine read something as a word the way is by counting the number of white spaces like the spaces between eh string of symbols then from there it can find the count How many words are in a novel in a poem and so on well from there then represented Machine into one eh calculation yes one Excel column like that how many times it appears and so on well
10:08 digital Hum is an effort to develop this intuition into a comprehensive study of literary works or literary phenomena of course the flow can be seen here there is first of course literature review resetap and so on this eh determination of researchap Ehan this research can be done for example with bibliometric analysis which is the application of digital Hum in library science maybe you have all received a seminar or tutorial about using for viewer like that, right Well that can be used to work in this context to identify the research gap where the literature review is like What is it like assisted by tools like that Then after that we formulate the concept in
10:53 our research question, for example we want to find the meaning of the word malam in the corpus of modern Indonesian poetry in the last 100 years, for example, the word malam appears in what sense, what are the changes like from period to period, then we operationalize the meaning we mean by this meaning here, what is the meaning here interpreted as, for example, the frequency of word occurrence, for example, like that or literary style, literary style is operationalized as what is the frequency of word classes and their arrangement, for example, like that Well then we operationalize each concept in the research question so that with that we can search for the data, acquire textual data and build our Kpus, for example, the Kpus from social media, yes, we scrape from
11:39 Twitter or Instagram or whatever, then we clean the data, label it and so on, then we go into how we organize a search for the data that can answer the research question like that, for example, using VOAN to check changes in word frequency from year to year and so on or running a text classification software called people and so on like that, more or less, so this is the general flow of digital humor research. Well, for example, what concrete examples are done by his friends at the Stanford literary lab, one of them is Sosi Sastraw, yes, he wrote a book called Horizon, he didn't read data from scanning 93,000 English novels that had been
12:24 reviewed by the government, or the Hatiust Library foundation, then he asked a simple question, how do English novels, which are on average from the 19th century to the 20th century, eh, depict women, so this is a classic question in cultural studies about gender representation in novels, yes, but the way he answered the question is interesting, yes, he looked from there, then he looked for what words describe women in novels, for example, he looked at it from the perspective of, ee, what is the heart, mind, ee, mind, and EE, lust, yes, these three words are more associated with women or
13:10 men in the 93,000 English novels that exist from 1800 to 2010. 2000 is like that, yeah, from there, then look at it, yeah, in the graph on the left, it looks like if it's above zero, it means the y-axis is above no. 0, then it's associated with women, while if it's below zero, it's associated with men. Well, we see that in 1800, heart or feelings are closer to women, yes, because it's above zero, yes, so in English novels, it's closer to women, while passion is closer to men because it's below zero, but once we enter the 1950s to the
13:55 2000s, both can no longer be distinguished, yes, because it's close to zero, both heart or passion, heart or lust, this vocabulary is associated with male or female characters, meaning, eh, it can no longer be sorted, yes, gender signals in the novel become increasingly difficult to sort out after the years - that means a change in the trend of depicting society about women and men, likewise with eh, another problem on the right, there are eyes, hair, chest, and wallets, wallets below zero because they are associated with men, it seems like it's still the same, yes, men as Bread Winner and so on But some others also experienced changes, yes, the eyes were
14:41 downgraded, women in the 19th year were always described first through their eyes, for example, like that, then it started to go down, replaced by, for example, hair. Yes, from 1950 onwards, it was more often like that. Well, this is an interesting example of how a gender representation can be read quantitatively. As long as we design the operational framework well, how do we calculate gender?
