A Neighbour’s Tree like Totoro’s
On the street side of the house, it almost looks like on the very, very hot days, when everyone avoids setting foot on the hot asphalt. It is as if all life was reserved for the garden side. A large part of Le Mans urbanism dates back to the beginning of the 20th century, with series of small town houses, most of them with gardens to the back side of the house. In many areas like ours, while houses close ranks to one another to the street side, it is the gardens that communicate to the back. And so we hear the neighbours gardening, we smell them barbecuing, we hear the swings of those with toddlers, and it feels anything but isolated and lonely. Separated by our man-high stone walls, we have the comfort of being on our own without feeling cut out of the world.
We have a small garden, but a magnificent view upon our neighbours’ larger gardens. I especially like one big, big tree, the view of which has a major soothing effect, especially when it has leaves. Two days ago, the first buds appeared, much to my delight. It has something of Totoro’s Tree to me.
So much for life in isolation: with Totoro guarding us, nothing bad can happen.
My plan to teach Greek to the kids is partly gaining traction. I have been pointed to great resources (in French) for beginners in Ancient Greek and will explore them with my daughter. Here are videos on how to write the letters properly, and here a handbook for teenage Ancient Greek beginners. I am already enjoying this a lot.
I have finally managed to actually finish a review yesterday evening. I could have postponed it, really, but the feeling of not getting anything done was so nerve-wrecking that I felt the impulse to at least finish one thing. And see: I started the day on a more optimistic note!
Also, the pre-print of a paper I contributed to is now online on my favourite archive HAL! You can find it here. It presents experiments I have been working on with ML colleagues from the Technical University in Berlin and more specifically with the main hero of this story, David Lassner (who also posted this here a couple of days ago). The goal of the paper is to explore automatic methods to evaluate censorship mechanisms (which we call alterations) in handwritten documents. We explore ways to cluster corrections to the manuscripts according to the type of correction and the motivation for them. We experimented based on my digital edition, Letters and Texts, but beyond that, the goal is to make the method reusable in other use cases. My ML colleagues go through great deal of methodological pedagogy in this paper, which I am extremely grateful for.
Writing it took us several years: a time to get to know, among others, the quality standards of our respective research areas. The difficulty to “talk the same language” when coming from such different areas as literary studies and informatics has been discussed at length in the DH community over the past years. But there is one thing in particular that only struck me very recently in this cooperation, namely when it came to publishing results and datasets. It is only very recently that I realized that one of the key challenges for SSH scholars who want to make their research digitally sustainable, e.g. archiving the data in a stable environment and in a standardized format, is a non-problem for MLers, who consume data more than they produce it.
Not wanting to leave a mess behind, but rather something that the following generations can build upon, has many dimensions: this one really matters a lot to me. I still have a few interesting conversations ahead of me with my ML colleagues… and certainly more papers to come, as I enjoy the explorative approach to literary data a lot.