I consider two specific NLP techniques for education research: topic modeling and word embeddings. I provide overviews of these techniques in the following section. I group this education research into three categories: Text as Observational Data, Automated Evaluation, and Adaptive Pedagogy. I also consider whether this work uses NLP methods to replace or supplement other existing techniques.
I do not describe the technical implementations for these NLP techniques in existing specific NLP tools performing these techniques. The state-of-the-art implementations for NLP techniques are rapidly evolving, and selecting a particular tool depends on researcher experience, objective, and data. Instead, I focus on how the general principles of these techniques can be applied to different types of education and education studies work.
Topic models are a type of NLP technique targeted to discovering hidden patterns in text (Boyd-Graber, Hu, and Mimno 2017). Words that frequently appear together in documents are grouped together as a topic. Documents can then be characterized by the presence or absence of different topics. While some categories may be obviously recognizable by humans (e.g baseball, hockey, and basketball could be grouped into a category for “sports”), topic modeling can also systematically generate less intuitive groupings of words.
Instead of grouping together words into topics, word embeddings create vector-based (numerical) representations for words in text. These vectors can then be compared to one another so that the distance between words indicates relationship strength. Recent word embedding work such as BERT (Devlin et al. 2018) has improved generalizability and support for words that have different meanings in different contexts (e.g bat when discussing baseball vs. animals).
This category refers to research using NLP techniques to transform existing language data into quantitative values that can be evaluated in the context of other variables. Nelson et al. (2021) evaluate the strengths and weaknesses of different natural language processing methods to support hand-encodings of documents. While initial NLP coding is typically imperfect, this supportive approach improves the overall speed of coding documents.
Alvero et al. (2013) demonstrate a replacement approach when evaluating the correlation between essay content and style with household income and SAT scores. The authors develop quantitative representations of essay content using topic modeling and other NLP techniques that are then used to predict family income. While both essays and SAT scores correlate with income, the authors find that essays are more strongly correlated than SAT scores. This suggests that efforts toward equity may place undue emphasis on scrutinizing quantitative evaluation measures when more intractable methods reproduce similar inequality, simply because quantitative values like SAT scores are more easily measured. Operationalizing non-numerical parts of educational processes such as college applications to draw these conclusions may be infeasible without NLP.
One of the earliest uses of computers in education was for grading essays (Page 1968). Existing systems use NLP techniques to evaluate how closely student essays are able to match keywords and their synonyms. Recent work in this field (Rokade et al. 2018; Wang, Chang, and Li 2008) aims to better evaluate essay structure and capture the semantic meaning of student essays using more sophisticated natural language processing tools. While this work largely focuses on developing fully-automated replacements for human grading, popular commercial tools such as Grammarly and Turnitin.com use NLP to provide a supportive tool where humans have the final say. For a deeper review of natural language processing in grading systems, see Rokade et al. (2018).
I am working with researchers at the University of Maryland to develop a flashcard recommendation system using word embeddings to establish semantic relations between flashcard content: KAR³L (Shu, Feng, and Boyd-Graber 2021). We believe incorporating these methods allows the system to better infer student knowledge on related topics, as well as better model the behaviors of human memory observed in psychology research (Ebbinghaus 1913; Erdelyi 2010). Research like ours suggests that NLP could improve learning efficiency by using better memory models to implement study methods recommended by pedagogicial research (Dunlosky et al. 2013).
Natural language processing systems for adaptive pedagogy may prioritize aims other than learning efficiency. Ruan et al. (2019) develop an adaptive chatbot named QuizBot that teaches and tests factual knowledge. Students learn more material through this medium than a traditional flashcard app using the same scheduling algorithm for items. While this system means students take more time to learn, they are also more likely to spend time using the app. This work highlights how NLP can be employed not just for efficiency but also for greater engagement.
Work in adaptive pedagogy could be used in both to support or replace existing forms of education. These two highlighted works focus on helping students acquire and/or retain fact-based information, which teachers could use in support of other strategies to apply what is learned. However, students may also use these applications to independently learn information suited to their interests.
There is ample room for NLP in education work. While I’ve focused on word embeddings and topic models, other NLP techniques like sentiment analysis and summarization can also be useful for work in all three of these broad categories: 1) Text as Quantitative Observational Data, 2) Automated Evaluation Systems, and 3) Adaptive Pedagogy. As demonstrated by this existing research, NLP acts as another methodological tool for achieving educational goals.
In projects bridging NLP with education, however, we should consider how our methods help us answer our research questions. Do our data actually help us answer our questions? What is the data source and are there ethical concerns about how to handle or interpret the data? Who are we including, and who are we excluding in our work? If you’re interested in conducting this research or learning more about how to handle these practical and ethical questions, I recommend Matthew Salganik’s online textbook Bit by Bit: Social Research for the Digital Age.
Abebe, Rediet, Solon Barocas, Jon Kleinberg, Karen Levy, Manish Raghavan, and David G. Robinson. 2020. “Roles for Computing in Social Change.” Pp. 252–60 in Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency. Barcelona Spain: ACM.
Alvero, AJ, Sonia Giebel, Ben Gebre-Medhin, anthony lising antonio, Mitchell L. Stevens, and Benjamin W. Domingue. 2021. “Essay Content and Style Are Strongly Related to Household Income and SAT Scores: Evidence from 60,000 Undergraduate Applications.” Science Advances 7(42):eabi9031. doi: [10.1126/sciadv.abi9031]{.underline}.
Ebbinghaus, Hermann. 1913. Memory: A Contribution to Experimental Psychology. New York: Teachers College Press.
Erdelyi, Matthew Hugh, and Jeff Kleinbard. 1978. “Has Ebbinghaus Decayed with Time? The Growth of Recall (Hypermnesia) over Days.” Journal of Experimental Psychology: Human Learning and Memory 4(4):275–89. doi: [10.1037/0278-7393.4.4.275]{.underline}.
Nelson, Laura K., Derek Burk, Marcel Knudsen, and Leslie McCall. 2021. “The Future of Coding: A Comparison of Hand-Coding and Three Types of Computer-Assisted Text Analysis Methods.” Sociological Methods & Research 50(1):202–37. doi: [10.1177/0049124118769114]{.underline}.
Ruan, Sherry, Liwei Jiang, Justin Xu, Bryce Joe-Kun Tham, Zhengneng Qiu, Yeshuang Zhu, Elizabeth L. Murnane, Emma Brunskill, and James A. Landay.
Salganik, Matthew J. 2017. Bit by Bit: Social Research in the Digital Age. Illustrated Edition. Princeton: Princeton University Press.
Shu, Matthew, Shi Feng, and Jordan Boyd-Graber. 2021. “Spaced Repetition Meets Representation Learning.” EACL 2021 HCI-NLP Workshop.
]]>I’ve read many new books in the past few years influential in shaping my thoughts1—at least while I was reading them. A few months after finishing a book, I typically remember little beyond the gist and at most a few specific takeaways2. When I wanted to reference Richard Rodriguez’s Hunger of Memory for a college essay, I found myself rereading the whole book. While I believe rereading can be an excellent use of time3, it’s not the most efficient method, and I’ve often wished to develop a better system for capturing key ideas and my thoughts about them. Unfortunately, my previous attempts at a ‘better’ system have always failed because I’ve habitually neglected maintainability. With time, I regress to highlights and brief notes within the book that I rarely revisit4.
Spurred on by discussion of note-taking in CGPGrey and Myke Hurley’s Cortex podcast, I’ve read Dr. Sonke Ahrens’ How To Take Smart Notes and have begun a zettelkasten system in Bear. I’ve found the guidelines laid out by Dr. Ahrens useful and this iteration of my notes system has been the best I’ve ever had5, but I’ve still had some problems with maintainability.
I’m currently retooling this system and catching up on a large pile of unsorted “temporary” notes. Part of this process has been identifying what notes have seemed out-of-place in the current iteration of my system, and what roadblocks are causing this build-up of notes in my inbox. Some problems are just related to how I’ve structured my notes in Bear, and something I might elaborate on in the future is the idea of “snippet notes,” which I’ve been storing haphazardly or not at all 6. The main problem I’ve found though is that note-taking doesn’t improve my memory, and while the idea of building a second brain is excellent, I still find it easier to see connections between ideas when my first brain remembers more.
I’ve found a zettelkasten-style note-system reduces the friction involved in getting started with writing and I like how it helps me remember how I’ve formulated certain thoughts, but the notes themselves have been no more helpful than highlights for retaining knowledge about the plethora of concepts, facts, and vocabulary introduced when reading informational texts. A zettelkasten is great for capturing and relating nuggets of original or captured ideas synthesizable into longer thoughts 7, but for the task of gaining an initial knowledge base to work from, flashcards seem superior.
I recently began revisiting the idea of employing spaced repetition and flashcards in my own life. While I maintained a 500+ day-long review streak in the flashcard app Anki to study for quizbowl in my sophomore and junior years of high school, I burnt out in the summer of 2019 and studied only intermittently in my senior year. That summer, I decided instead to work on improving flashcards and spaced repetition algorithms through machine learning, collaborating with other researchers at the University of Maryland to create the flashcard app KAR³L for studying trivia. We’ve recently released a progress update from our first phase of experimentation and plan to transition into a new phase soon. So if you’re interested in both helping our research and studying for Jeopardy!, quizbowl, or other knowledge-based tournaments, now is a good time to try out KAR³L and be notified when we roll out our upcoming updates8.
I began using flashcards again personally after realizing how they could complement note-taking. My largest problem has been remembering old highlighted thoughts and ideas, and spaced repetition flashcards offer a potentially maintainable solution for resurfacing them. After deleting my old 20K+ flashcard collection in Anki a few months ago 9, I’ve begun using Anki to store key information from content I consume 10. In particular, I’ve been creating flashcards from textbooks like Matthew Salagnik’s Bit By Bit, Goodfellow et. al’s Deep Learning, Annette Lareau’s Unequal Childhoods, as well as various papers and articles. I’ve been using flashcards to remember terms and concepts that could come in handy in the future. Recently, I’ve also begun creating ‘quote’ flashcards, which works similarly to Readwise, resurfacing content that previously would have highlighted and forgotten. Both of these flashcard types provide our brains more material to work with when writing, facilitating the process of discovering connections and references in both our notes and essays.
While I enjoy reviewing flashcards in the AnkiMobile iOS app, I don’t like adding new flashcards with it. I miss features available in desktop Anki through add-ons like Frozen Fields and GODMODE: faster shortcuts and cloze switching. I usually find AnkiMobile’s preservation of formatting more a hindrance than a help, and copying text from pdfs has always been tedious due to the weird formatting issues frequently present. I also want to quickly add cards through an action extension and have a place to park material for later use. As I create most flashcards on my iPad, where I do most of my reading, I’ve chosen to address my needs by taking advantage of AnkiMobile’s URL scheme support to develop an app specifically for creating flashcards. This project has taken longer than expected, but the flexibility afforded by my gap year has helped me see it through to release on the App Store (for free!).
I’ve recorded a playlist of tutorials demonstrating various features in Flashcard Adder available here. Here’s a quick list of certain useful features available in Flashcard Adder not currently available in AnkiMobile, with some associated video tutorial links.
I’m not aiming to replace AnkiMobile’s built-in Add Note screen for all use-cases with this app. Instead, I optimize for features that make creating flashcards on mobile easier for myself, and hopefully others as well. Features in the built-in adder that will likely never be supported in Flashcard Adder include attributed text, images/drawings/audio, and tags11. But while lacking these features cuts down on the possible types of flashcards, Flashcard Adder facilitates more maintainable workflows for adding flashcards through features like the action extension and dynamic note type switching.
During the first half of quarantine, I brushed up on Swift fundamentals by going through Angela Yu’s Udemy iOS development source and reading Paul Hudson’s Swift Design Patterns and Pro Swift books. I finally got to put some of what I learned to practice with this app, and I also learned more about generics, building UI programmatically, and working with the new APIs for UICollectionView. Unlike with my first published app, QuizDB Mobile, I was more conscious of app architecture and strived to avoid certain mistakes that made that previous app harder to maintain. Despite this, the structure of Flashcard Adder is not pristine and I’ve likely still made some rookie mistakes. I’d love to receive comments from more experienced iOS app developers about how I can improve this code.
Source code for Flashcard Adder can be found on Github. If you find this project interesting, please give it a star!
Somewhat surprisingly, given the general ineffectiveness I’ve come to associate with Youtube ‘productivity’ videos, John Fish’s Reading a Book a Week is Changing My Life video was highly influential in pushing me to develop this habit. ↩
Andy Matuschak wrote a long essay about this issue titled Why books don’t work that I remember being interesting. It’s been a while since I read this and I didn’t take notes so unfortunately, I don’t remember much anymore (heh), but I believe Matuschak proposes developing new mediums for communication. Like what Cal Newport notes about personal task management in The Rise and Fall of Getting Things Done, the systems we develop are only bandaged fixes for wider structural problems. But while widespread change is ideal, workarounds like the flashcard system outlined may just be the best we can do for now. ↩
I’ve been reliving my childhood through John Flanagan’s Ranger’s Apprentice series these past few weeks. ↩
The lack of a standard, open annotation standard in the epub ebook format doesn’t help either. Marvin 3 began crashing after I updated to iPadOS 14.2, and I spent hours exporting my annotations out of the app after rolling back the update. Even then, I was only able to export to CSV format and Marvin 3 on my iPhone, as no other epub reader would be able to display these annotations inside the ebook anymore. Given the nature of my old highlights, I’m uncertain if these out-of-context highlights would even make sense at all. ↩
Perhaps this doesn’t necessarily mean much, since all my previous ‘systems’ have been haphazardly unorganized, intermittently used, and spread out across an eclectic collection of apps. ↩
To briefly expand on the idea of snippet notes, think of code snippets and text that might go into a program like TextExpander. These notes store fully developed sentences/paragraphs that I feel may have the potential to be recycled in a future use with at most slight variations. In my current makeshift-system in Bear, I’ve set up a certain repository for application essays I’ve written and have begun storing variations of my responses to an answer in the same note. ↩
Dr. Ahrens advises in Chapter 5 of How to Take Smart Notes that one should be mindful of how your notes can be used in the future for writing. ↩
You can follow development progress on our Github repository. ↩
While the principles of the KAR³L scheduler should work with language learning, there are some unique considerations we haven’t figured out yet. KAR³L is currently evaluating the performance of different schedulers in the area of trivia and therefore using it for my personal aims risks diluting our current research focus. For this reason and a few others, I elected to stick with Anki instead of KAR³L, even though I believe our scheduling algorithm is superior. ↩
Unrelated to note-taking, I’ve also started using an amazing Chinese flashcard deck made by Timo Horstschafer to finally learn how to read Chinese. ↩
To me, tags are generally an unnecessary distraction. I used to religiously tag almost every note before I realized how doing so slowed me down without providing much benefit since I rarely used them afterward. Perhaps I’ll change my stance on tags in the future, but that’s why I haven’t added support yet. Implementing a simple tag field shouldn’t be that difficult though, and if someone wants to submit a pull request doing so, I’m happy to include it. ↩
Embracing OnePlus’ motto of “Never Settle,” I’ve constantly sought out new ideas or apps in a quest for improvements to my quality of life. When I do decide on an app or a system, it’s usually not long before I find something frustrating about it. Sometimes this works out and I find useful ideas for personal projects, like Mindful.ly, QuizDB for iOS, Flashcard Adder, and an in-the-works e-book reader app. Other times, I end up spending more time looking for nonexistent better alternatives instead of working with what I have.
I’ve been trying to work on this since the start of quarantine and in particular since my gap year ‘officially’ began in August. While I’m still giving myself room to explore and try out new things (after all, sometimes shiny new toys can prove useful), I’ve started focusing on developing actual workflows within these tools to achieve my goals. The systems I implement and the ways I improve on them over time are more important than the apps I choose, but they must be refined to play well with the available software.
I used to spend an inordinately long time implementing my desired workflows through inefficient workarounds. Back when I used Todoist, I implemented a habit tracker with Habitist, which involved learning Heroku app deployment 1. My largest barrier towards actually getting things done ended up often being my preconceptions for how to get things done. After putting in the time to set up whole new apps or systems for task management, it wasn’t long before whatever new fragile system I built fell apart under the strain of a busy stretch. Things usually broke down into completing assignments the day before they were due and then spending my free time working on whatever one side project was occupying the most prominent amount of space in my mind. Everything else fell on the back-burner, and the friction often involved in adding new tasks I sometimes avoided adding them. I spent more time setting up my systems than actually using them.
When my carefully crafted systems fell apart, I sometimes reverted to writing down my most important tasks and goals on paper. What’s great about this analog system is that its limitations force me to rethink how to structure my workflow to work with it. There are many inconveniences to deal with in a paper-based system solvable in a digital app, but the challenges of paper make it easier to quickly create something meeting my immediate needs. With digital apps, everything feels like it should be possible, so when I tried to make my digital system do more than paper, I’ve often found my tinkering just made things worse.
In part, these failures suggested that my notions for how to get things done could use work. So, a few months ago, I read David Allen’s aptly named book Getting Things Done. Like most self-help books, it’s quite possible to boil down the main ideas from the book into a relatively short article, and of course, there are many of those out there already. If you have the time though, I think there still is value in giving the real book a try—I find it generally easier to absorb ideas from a book than from condensed summaries. I won’t summarize the book here, but I’ve incorporated many of Allen’s ideas into my workflow and do recommend reading at least the first half of the book.
When I redid my system after reading Getting Things Done and some other guides, I chose to stick with Things 3 as that was my task management app I was already using 2. It’s a beautiful app with design choices inspired by the Getting Things Done system, and Cultured Code (the developer of Things) provides a useful introductory guide for getting started that’ll put you in a decent place productivity-wise. Personally, however, not all of their advice stuck when I first read it, so I doubt a little informational redundancy will hurt here. In this post, I’ll provide a more detailed look at how I’ve personally worked with Things and a few helper apps to develop a task management system I’ve been happily using these past few months.
Like all task management apps, the basic building block in Things is a task (AKA a To-Do). To-Dos are generally discrete, actionable items completable in a single session of time. Projects contain a list of actionable tasks that, when completed, typically result in an end accomplishment. Tasks and projects can be grouped under an area as well as labeled and filtered with an arbitrary number of user-specified tags. Within a project, headers can optionally group different tasks to provide visual separation. A Project and its tasks typically aren’t completable in a single session. This contrasts with subtasks for a task, which represent minor steps or checks for completing a task. However, this difference isn’t sharply defined and whether you choose to employ a project with tasks or a task with subtasks is up to you, but I encourage you not to be hesitant about having a lot of projects. I personally rarely use subtasks—only as a checklist of things I check to make sure I covered everything I needed to for that task.
It can be tempting to think that this makes for a folder-like hierarchy of Area -> Project -> To-Do -> Checklist/Subtasks. When I began using Things, this is what I envisioned, but this categorization system is not strictly hierarchical. Nearly all my tasks don’t contain subtasks, and I have a sizable number of tasks (e.g “Fix Spotify local files on iPad”) within an Area rather than within a particular Project. I occasionally also have projects and tasks that don’t fall into any Area.
For a concrete example of these differences, my Personal area has the project “Productivity System Blog Post” with the tasks “write first draft,” “edit draft,” and “upload to site.” Within the “upload to site” task is a list of subtasks I’ll check to make sure I cover before I post. Inside Personal, I also have singular tasks like “Fix Spotify local file play on iPad.” I also have a few recurring tasks in Personal like “Conduct weekly review” and “Review book list in Calibre.”
In addition to this organizational system, Things has the built-in lists of Inbox, Today, Upcoming, Anytime, and Someday. The Inbox is a list of unsorted items, Today contains items that have been assigned a start date or deadline of that day, Upcoming shows a list of items with upcoming start dates or deadlines, Anytime contains items without a specified due date and not placed into Someday, and lastly Someday contains To-Dos and Projects that aren’t likely or able to be worked on at the current time.
For the most part, these built-in lists are pretty self-explanatory, but it’s easy for the lines to become blurred and cause problems. The Inbox might grow into a large dump of way too many unsorted tasks, or it could be underutilized/not utilized at all. If there are too many tasks in the Today view and tasks are frequently left incomplete for the day, Today becomes essentially a higher priority Anytime list. If there is an overload of tasks in Anytime, it can effectively become just another Someday list. In turn, Someday can grow into an archive of tasks and projects that will likely never be completed but that one doesn’t have the heart to trash. Similar problems can arise when working with Areas, Projects, To-Dos, Headers, Tags, and Subtasks. In particular, it’s hard to strike a balance between too few and too many tags. In the new few examples, I’ll go through times when I’ve encountered issues like these during my time using Things and how I’ve adjusted over time to create a more cohesive system.
While in high school, I could never really figure out how to properly organize homework and other class-related tasks. For a while, I kept a project for each class. Almost all homework, except for extremely long-term projects, was stored a singular task sorted under the corresponding class ‘Project’. Later on, I wanted to have Headers in my Areas for each class, as I was reluctant to take the step of creating wholly separate areas for each class. As many classes assigned work infrequently, I was often left with empty Projects taking up space on my sidebar. I struggled to always add tasks as well, and when I did I typically bypassed the Inbox step. My Today view also often featured tasks I hoped rather than needed to do that day, leading to many tasks carrying over day after day as I failed to complete them. For a time, I started using tags to label different classes but grew annoyed by the additional steps it took to do so.
My problems in these circumstances largely stemmed from my attempts to use Projects similarly to folders, as a tool primary for organization. When I began using headers, I was reluctant to make projects for homework, as to me that would break my precious organization. Looking back now, I think there was little use in my preconceived notion that all my class-related tasks needed to be organized under a label for that specific class. After all, organizing them on aggregate with the School area would have been fine; there wasn’t much point differentiating when each class typically had no more than a few tasks/projects going on at once (and some rarely ever had any). Without these sometimes empty projects taking up space, the sidebar would only actual projects containing actionable next-actions.
Solving the problems I faced with the Today view and missing tasks requires a little more thinking and another change in workflow. When homework was assigned, I always immediately tried to add the task in its proper place, going through the slightly cumbersome task of adding a to-do to what I thought was the correct organizational ‘folder’/project. This meant that when I didn’t feel like there was enough time at hand to put the task where it was supposed to be, I sometimes just didn’t add the task. What I should have done more often, however, was use the Inbox list as a parking place for quick notes about tasks. With this as a routine, I would have been less likely to skip and then later forget to add tasks. For simple homework, it might also have been easy to just do it later while cleaning up the Inbox list. If I used the Inbox, this would also allow for more deliberation about whether that homework could benefit from being split up into a project or by being assigned certain tags. For example, I think it would have made sense to make math homework into a project with each problem or chunk of problems as a separate task since for me math typically took up the largest amount of time. If tasks were chunked up in this manner, I would be more likely to complete everything in my Today view, and then be able to move on to my Anytime list with my free time. With an Inbox, I’ve also found it far more fun to go through the process of doing the little tasks and organizing the rest.
Since the start of quarantine, and particularly since summer began, most of my tasks are things I’ve personally chosen for myself. The changes in my approach to projects, the Inbox list, as well as the Today and Anytime views, have proven helpful in these circumstances as well, but this period of self-direction has brought to light other issues in my workflow arising from lack of deadlines and the increase of potential things to do (as well as time to do them).
My Someday list has historically been where tasks, projects, and ideas go to die. However, I’ve recently realized how much more useful this list could be. My Someday list now only contains projects and tasks I can see myself doing some time in the next year. To satisfy my need to have an archive of other tasks and wishes for the future, I’ve started to instead store these ideas and thoughts in Bear, my notes app. This doesn’t mean that I’ll do everything in my Someday list within a year (in fact, I definitely won’t be), but keeping this in mind reduces the informational overload in the list. This in turn makes it more attractive for me to browse through every few weeks or when I’m low on tasks and projects currently active in Anytime, Upcoming, and Today. Since Someday is no longer a graveyard, I’ve also become more comfortable putting tasks or projects that require some variable length of waiting before I can continue work, usually due to sequential or collaborative tasks where I can’t complete the task or continue work on the project until whatever roadblock has been cleared.
The importance of having a system for conducting ‘reviews’ has been hammered into me quite frequently, but I’ve always found it difficult to follow through with this task. Now that I’m in a scenario where I’m not overloaded with tasks in the Today view and have regular movement between the Inbox, Today, Anytime, and Someday views though, I’ve found it more natural to complete these reviews. I’ve found going through these tasks to be a relatively enjoyable activity to work on while on a walk or when I’m too lazy to do other work. Since I’ve finally experience how this organizational shuffling is actually useful and somewhat fun to do, this habit has become far easier to keep.
The last major feature of Things I’ve begun employing are tags. Tags have been tricky to get right in the past, and I’ve usually avoided using them at all save for the few misguided times I spent hours retroactively adding tags I later never used. I’ve always found the idea of adding contextual information to tasks intriguing, but I usually went overboard developing an unnecessarily elaborate tagging system. I used to also find adding tags particularly cumbersome in my previous workflows, as they added valuable time to the process of quickly adding new tasks to the correct place in a system.
Since I started using the Inbox, however, labeling with tags has become just another quick step during the Inbox review process when I decide not to immediately work on the task or project and instead file it into the correct place in my organizational system. I’m still continuing to figure out what tags are useful though. Since I began using tags again, I’ve typically labelled tasks and projects with three label types: estimated time for completion (<15 min, 15-30 min, <1 hour, 1-2 hours, long, multitask), tools needed for completing the task (Mac, iPad, iPhone, none), and intensity of task (low, medium, high). I’ve sprinkled in other tags but aside from the waiting tag I’ve generally found them unhelpful. Recently, I’ve decided to stop using the intensity label for now since my intensity estimates tend to be quite arbitrary and when I look back I often disagree with my initial assessments. It’s also usually quick easy to pick something to work on after just filtering with tool and time constraints.
While it’s crucial to recognize that projects represent a list of actionable tasks that will eventually be completed, I’ve personally found it helpful to have two exceptions to the general rule: Store and Review Again. Both of these projects are never-ending and serve essentially as organizational folders, resembling my initial view of what projects are for. However, since there’s almost always many tasks in these folder projects, I don’t find them to be a waste of space the way my school class folder projects were.
I use Calibre and Zotero to store everything I read nowadays, and I love these programs. Unfortunately, both are desktop-only apps 3. When I review the tasks in my Inbox, I don’t always feel like immediately completing these storage/acquisition tasks when I’m on my Mac, and it’s simply impossible to complete when I’m on my iPad or phone. As a result, Store is where I keep links to texts I need to store later. I tag this project with ‘mac’ so they appear on my Anytime view filter for when I’m at my computer. I prefer to bunch up these tasks as doing all of these tasks at once as the repetition puts me into a certain relaxing zen state and I feel likely also saves a sliver of time.
Review Again is a more tenuous addition to my system that currently stems from needing a place to store articles and papers that I want to review another time on my computer or tablet to develop notes and ideas. This list has gotten quite long though, so perhaps there’s a better way to address this situation.
I don’t typically use Things for habits I want to do every day—the one exception being a recurring “Walk outside” project contains a list of items I should bring 4. Instead, I track my habits in a simple spreadsheet with a row for each day and a column for each habit. I’ve tried the free tiers of various habit tracking apps, but the restrictions and advertising are annoying and a paid app doesn’t seem justifiable for what is essentially just a nicer looking spreadsheet. While Google Sheets isn’t great on mobile, but it’s not like I have to do much with the spreadsheet each day.
For tasks that need to be done by or at a certain time, I’ve been using Due for the past few years to nag me into actually completing them. This app is wonderfully simple—set a reminder at a certain time for a task and Due will repeatedly notify you about the task until you either complete, reschedule, or delete it. For example, I have reminders to reach out to people, check in on scholarships, and cancel subscriptions at certain (sometimes arbitrarily set) times. I also have recurring reminders for things like “Back up database” 5. Even when my regular task management systems broke apart, I’ve always kept this app handy to nag me whenever I need it.
People often talk about how they can’t switch away from Apple because they’re too closely tied to the ‘ecosystem’. This is the type of harmony I hope to reach with the software I use with my workflows. Apple regularly updates its products with improvements, and I similarly expect make changes to my workflow when circumstances change, or when I review what’s working and what’s not. But just as how making a change like switching from iOS and Android involves high transition costs and careful decision making, I aim to treat jumping ship from my established systems and software in a similar manner.
I’m currently in the process of applying this principle of standardization to other aspects of my work and life as well.
Five years from now, I hope I can look back and not see many changes to this list, with the exception of my in-development ePub ebook reader of course.
This remains my only experience with Heroku. It was also my first experience with the mess that is deployment ↩
Apple revamped Reminders in iOS 13 and macOS Catalina, and had I not already bought Things 3, I likely would have tried coming up with a system using that app instead. My system in Things 3 works well for me, and I don’t expect to move away from it anytime soon, but Reminders is a solid app to keep in mind if you don’t want to pay. While I don’t believe my system transfers easily into Reminders, I don’t think it really should. If you start with Reminders, I’d spend some time looking at how other people use it and then experiment with it yourself to see how you can develop a system around that app. ↩
There are some clients for Zotero on iOS that I don’t use. I prefer using Zotfile and iCloud Drive. ↩
This could probably have been a task with a checklist, but I appreciate the visual separation. When I don’t walk outside, I cancel the project for the day. ↩
I know something like a cron job might simplify this further, but I haven’t spent the time learning to develop something that works and Due nagging me works well enough for now. ↩
相关信件: 致艾琳 (Eileen), 一涛 (Yitao), 朱锟 (Kalos)
致艾琳、一涛、朱锟(Kalos)和全美国华人,
我是束骏杰,一名耶鲁大学新生。 艾琳,读了你的公开信后,我更坚定的相信选择耶鲁是正确的,我很期待将来与你合作以实现我们所期望的改变。 但是,你的公开信中的部分观点我无法赞同。我担心这些观点会在美国华裔乃至亚裔群体中引发分歧,导致大家无法达成共识并采取行动。在阅读了一涛和朱锟(Kalos)的深入答复后,我也希望表达一些我的想法。
艾琳,你说“我们今天所拥有的一切都归功于黑人的努力”,我不认同这种观点。这种说法是在淡化美国华裔在平权斗争中付出的努力。 我们都知道,现今在美国出生的婴儿可以通过“落地国籍”原则自动获得美国国籍,但人们往往不了解黃金德(Wong Kim Ark)为了巩固这一权利所付出的巨大努力。 类似的,在现今美国的公立学校中英语作为第二语言的教学已经十分普遍,但人们常常忽视在Lau v Nichols案中的美国亚裔的贡献。 围绕种族主义的斗争不管在过去还是将来,都不仅仅是黑人与白人之间的斗争。反种族主义斗争跟所有族裔都息息相关,需要所有人共同合作。 如你所说,文森特·钱(Vincent Chin)谋杀案引发的抗议活动正是这种跨种族合作的重要代表,但是这一符号恐怕已经被众多美籍华人—遑论其他族裔—所遗忘了。
在现今意识形态极端分裂的美国,我们见证了因种族主义带来的种种恶行:乔治·弗洛伊德(George Floyd)谋杀案、 因新冠引发的亚裔仇恨,不一而足。在这种环境中保持缄默、拒绝发声、进而助长种族主义,是错误的。很多人都犯了这样的错误,这也包括很多美国华人。我们应该反思、批评我们之中保持沉默的人,但同时也要明确我们的立场。与其思考我们亏欠了什么,我们更应该把能量集中在为更好的美国做出贡献。
一涛,您指出针对美国华人的刻板印象是正确的。 但是,这种理解同时也在传播和加深对其他群体的刻板印象。在把美国华裔的成功完全归功于中国文化的同时,我们也在暗示其他的群体缺乏这样“优越”的文化。 这种思考方式在强调单一文化的优越性:因为我们的文化鼓励自力更生,鼓励努力工作和创业,所以美籍华裔便比其他群体要更优越。 但正如朱锟(Kalos)所说,这样我们不也在暗示其他美国人对这些价值观不抱有类似的自豪感? 若真是这样,为什么在西班牙,二代华人移民的学历却不如其他少数族裔[6]? 西班牙华裔是否缺乏所谓让美籍华人成功的中华文化价值观?
为了正确解释美籍华人与其他少数族裔在成就上的差异,我们必须了解我们的起点,族群资本和种族主义的影响。 艾琳提到了1965年的重要性:那一年移民政策的变化导致了对拥有高等教育的技术移民的偏爱。 虽然美籍华人在美国有一段漫长而动荡的历史,我们大多数人今天能来到这个国家,也是这个政策的结果。
一涛,我毫不怀疑您和我们所有的第一代华裔父母一样,都为实现美国梦而付出了极大的努力。 如您所说,许多一代移民在到达美国时几乎没有任何退路。 但必须指出的是,尽管缺乏物质资源,这些一代移民其实已经拥了许多人没有的宝贵资源:高等教育。 50%的美籍华裔移民拥有本科或更高学历,而在中国只有4%的人有相同的学历[5]。 同时,只有5%的墨西哥裔美国移民拥有本科学历或更高学历,而17%的墨西哥裔拥有同样的学历。 在全部美国人中,拥有本科或更高学历的也只有28%。鉴于我们作为一个社区将教育作为未来成果的重要指标,您难道不认为在移民美国过程中所经历的严格筛选是我们集体成功的重要因素吗?另外,由于美国华人整体的教育水平较高,我们也可以通过共享资源来更好的支持教育水平相对落后的华人家庭。
与美国不同,在西班牙,只有40%的中国移民从高中毕业,而拥有本科或更高学历的人不到5%[6]。 预计只有11.7%的二代中国西班牙人会获得大学学位,但在西班牙的所有少数族裔中这个数字是21.4%。 尽管来自相同的文化背景,西班牙华裔和美国华裔在教育水平上存在明显差异。由此可见,中国文化不足以解释美国华裔的成功。
当我们比较子女和父母的教育水平时,我们发现在美国第二代拉丁裔比起他们的父母的进步最大[3],但是这一上升趋势并没有延续到他们的子女身上。这种现象,即第1.5代和第二代美国移民在教育上取得的巨大成功,在社会学中可以用移民乐观主义从两个角度解释。 首先,对于那些自愿来到美国的一代移民,在适应的过程中遇到的种种问题更像是暂时的,他们更有意愿去找到解决方案并克服挑战。而另一个角度令人沮丧: 移民后裔在争取美国梦的过程中遇到的种族歧视使他们对这个梦失望了。
与美国华人一样,韩裔及朝鲜裔是在美国也被视为“模范少数族裔”。 但是在日本,韩国人和朝鲜人受到了类似黑人在美国受到的歧视。 在小说《柏青哥》中,李敏进(Min-Jin Lee)含蓄地展示了这种种族主义的破坏性影响[4]。书中是这样描述一个在日本的朝鲜孩子的挫败感的:
摩撒知道他要变成为一个坏朝鲜人了。警察经常以偷窃或在家私自酿酒的罪名逮捕朝鲜人。每个礼拜,他住的那条街上都有人被警察抓走。诺亚说,一些朝鲜人违反了法律,所有朝鲜人都会背上恶名。在亚野区的每一个街区都有男人打老婆,每一个街区都有女孩子在酒馆里工作、为了钱出卖肉体。诺亚说朝鲜人必须努力工作,变得更好,借此提高形象。而摩撒只想把那些说脏话的人臭揍一顿。在亚野区,有一些丑陋的老女人骂街,还有一些男人喝得酩酊大醉,睡在屋外。日本人不希望朝鲜人住在自己家附近,因为他们很脏,和猪住在一起,孩子身上还有虱子。此外,还有人说朝鲜人还不如贱民,因为贱民至少还有日本血统。诺亚对摩撒说,他以前的老师都说他是优秀的朝鲜人。而摩撒明白,由于自己的学习成绩差,又不懂礼貌,那些老师一定认为摩撒是个糟糕的朝鲜人。
他妈的那又怎么样? 如果其他十岁孩子认为他蠢,那也无所谓。如果他们认为他很暴力也无所谓。如果有必要的话,摩撒不怕把他们打得满地找牙。“你以为我是野兽”,摩撒想,“那么我就可以成为野兽,让你吃点苦头”。摩撒无意成为优秀的朝鲜人。那有什么意义?
在给艾琳和朱锟(Kalos)的回应里,很多人指出了这不仅是两代人之间的差异。在我这一代之中,许多人也持有不同的看法。 是的,面对种族问题拒绝发声的问题不局限于一代或几代人。我们不能一概而论的宣称我们的父母都是种族主义者,也不能自诩是摆脱了偏见的“受过教育的精英”、使命是帮助父母摆脱种族主义的偏见。事情恐怕没有这么简单。 当然,我们不能否认许多美国华裔对黑人的种族主义歧视。估计所有人都听说过、甚至使用过“黑鬼”,这个词的存在和传播的广度就足以说明了美国华裔群体中相当一部分人持有种族主义观点。
归纳总结与偏见之间的界限线模糊,我想我们每个人都曾越过这一界线。但是通过参与像这样的讨论和深入思考,我们大家可以共同努力,消除偏见,促进反种族主义思想,以创造我们所希望的更加平等、司法更加完善的、更好的美国。
奥巴马在2004年说,“不存在所谓黑人的美国、白人的美国、拉丁裔的美国、亚裔的美国 。 只有一个美利坚合众国。” 他说:我们共同的美国公民身份比我们之间的差异更为重要 [1]。 但是,乔治·弗洛伊德(George Floyd)谋杀案揭示的事实是,奥巴马对美国的描述并不是对现实的反映,而是对未来的展望。我们并不是第一次听到这样的愿望:亚布拉罕·林肯和马丁·路德·金都说过这样的话。改变需要时间,但是我也们也应该回想起40年前詹姆斯·鲍德温(James Baldwin)发出的质问[2]: “我还要等多久?”
一涛,我同意我们应该谴责骚乱带来的暴力和对小企业的破坏。 如您所说,这些行动“完全不利于解决问题。” 但是,我们要分清暴徒和示威者。 我们不能让暴动者的行动淹没了示威者的声音: 他们不愿继续忍耐了。 暴动者的行动说明了问题的严重性,但我们不应过度关注暴动而忽略了真正重要的东西:为警察改革和反种族主义做出贡献。
祝好,
束骏杰 (Matthew Shu)
翻译: Shi Feng, 徐文骅 (Brian Xu)
[1] B. Obama, “Barack Obama’s Remarks to the Democratic National Convention,” The New York Times, Jul. 27, 2004.
[2] K. Thorsen, James Baldwin: How Much Time Do You Want For Your “Progress?”
[3] G. Kao and M. Tienda, “Optimism and Achievement: The Educational Performance of Immigrant Youth,” Social Science Quarterly, vol. 76, no. 1, pp. 1–19, 1995.
[4] Min Jin Lee, Pachinko. Grand Central Publishing, 2017.
[5] J. Lee and M. Zhou, The Asian American Achievement Paradox. Russell Sage Foundation, 2015.
[6] J. Yiu, “Calibrated Ambitions: Low Educational Ambition as a Form of Strategic Adaptation Among Chinese Youth in Spain,” International Migration Review, vol. 47, no. 3, pp. 573–611, 2013, doi: 10.1111/imre.12037.
]]>Referenced letters: Eileen, Yitao, Kalos
To Eileen, Yitao, Kalos, and the Chinese American Community,
My name is Matthew Shu, and I am a rising freshman at Yale University. Eileen, your letter reaffirmed to me why I have chosen to attend Yale, and I hope to collaborate with you in the future to achieve the changes we both call for. However, I disagree with parts of your letter and I worry that similar disagreements have distracted others in the Chinese American and wider Asian American community from accepting your premise and call to action. After reading Yitao’s and Kalos’ thoughtful responses, I hope to add my voice to this discussion.
Eileen, I do not agree with the mindset that “we owe everything” to Black Americans. When we make this claim, we’ve discounted our own agency in the fight to make America a more equal society. We’ve overlooked how Wong Kim Ark’s struggle cemented for all Americans the right to birthright citizenship. We’ve forgotten how Asian Americans in Lau v Nichols fought for “English as a Second Language” programs to become widespread in America’s public schools. The struggle against racism is not and has never been the struggle of Black Americans on one side and White Americans on the other, with everyone else on the sidelines rooting for a particular side. As you accurately point out, the protests in the aftermath of Vincent Chin’s murder represent a powerful moment of interracial cooperation. But he is not remembered by most Americans. Asian Americans are not the only ones who have forgotten this symbol of cooperation.
In this divisive atmosphere, Asian Americans have not been the only ones complicit in this “kind of silence” in the aftermath of heinous racist acts—from George Floyd’s murder to COVID-19-fueled xenophobia. We must chastise those in our community who stay silent, but in remembering our past, we must also clarify our message. What we owe Black Americans and other protesters today is not a debt, but rather, further cooperation in movements to create a better America.
Yitao, you are correct in calling out stereotypes against Chinese Americans. However, I believe you have fallen victim to similar stereotypes about other communities, as well as our own. When we attribute the general success of Chinese Americans as a product of our Chinese culture, we are implying the inferiority of Black and Hispanic cultures. By holding this viewpoint, we suggest that Chinese Americans are superior to others because of the cultural values we hold of self-reliance, hard work, and entrepreneurship. But as Kalos notes, are we then suggesting that other Americans do not place similar pride in these values? If so, why do second-generation Chinese immigrants in Spain have lower educational attainment than all other Spanish minorities [6]? Are Chinese Spaniards somehow deficient in the Chinese cultural values you claim as the primary reason for Chinese American success?
To properly explain the disparities in achievement between Chinese Americans and other minorities, we must understand the effects of starting points, ethnic capital, and racism. Eileen mentions the importance of 1965, the year a change in immigration policy led to increased preference for highly skilled and educated immigrants. While Chinese Americans have had a long and tumultuous history in America, most of us today have come to the country as a result of this act.
Yitao, I do not doubt you and all of our first-generation Chinese American parents have worked extremely hard to achieve the American Dream. As you stated, many first-generation Chinese Americans had little to fall back on when they arrived in America. However, it is crucial to note that despite this lack of material resources, many first-generation Chinese Americans had a valuable resource: prior education. 50% of Chinese American immigrants have a Bachelor’s degree or higher while only 4% do in China [5]. Meanwhile, only 5% of Mexican American immigrants have a Bachelor’s degree or higher while 17% do in Mexico. In the general American population, only 28% have a Bachelor’s degree or higher. Given how much we as a community value education as an indicator for future outcomes, wouldn’t you agree this hyper-selectively among Chinese American immigrants is an important factor in our collective success? Because such a large proportion of our community is highly educated, we are also better able to support children from less educated families by sharing community resources.
Unlike in America, only 40% of Chinese immigrants in Spain have even graduated high school, and less than 5% have a Bachelor’s degree or higher [6]. Only 11.7% of second-generation Chinese Spaniards are expected to obtain a university degree, compared to 21.4% across all ethnic minorities in Spain. Despite sharing the same home culture, there is a clear discrepancy in educational attainment between Chinese Spaniards and Chinese Americans. Chinese culture is an inadequate explanation for Chinese American success.
When we measure success by the change in education level relative to parents, it is actually second-generation Hispanic Americans who are the most upwardly mobile [3]. Unfortunately, this upward trend does not continue to third-generation Hispanic Americans. Immigrant optimism is a hypothesis in sociology explaining this statistical observation that for some ethnic and racial groups, 1.5 and second-generation Americans achieve the greatest educational success. There are two reasons for this idea. The first is that voluntary immigrants to America see adjustment issues as temporary and are more willing to find solutions to overcome the challenges faced. The second, more depressing, reason is that later generations of minorities have become disillusioned by the racism they have encountered as they strive for the American Dream.
Like Chinese Americans, Korean Americans are another ethnic group in America seen as a “model minority.” But in Japan, Koreans are targeted similarly to Blacks in America. In her novel Pachinko, Min-Jin Lee poignantly shows the damaging effects of this racism [4]. Here is an excerpt capturing this frustration in a Korean-Japanese child:
Mozasu knew he was becoming one of the bad Koreans. Police officers often arrested Koreans for stealing or home brewing. Every week, someone on his street got in trouble with the police. Noa would say that because some Koreans broke the law, everyone got blamed. On every block in Ikaino, there was a man who beat his wife, and there were girls who worked in bars who were said to take money for favors. Noa said that Koreans had to raise themselves up by working harder and being better. Mozasu just wanted to hit everyone who said mean things. In Ikaino, there were homely old women who cussed and men who were so drunk that they slept outside their houses. The Japanese didn’t want Koreans to live near them, because they weren’t clean, they lived with pigs, and the children had lice. Also, Koreans were said to be even lower than burakumin because at least burakumin had Japanese blood. Noa told Mozasu that his former teachers had told him he was a good Korean, and Mozasu understood that with his own poor grades and bad manners, those same teachers would think Mozasu was a bad one.
So the ** what? If the other ten-year-olds thought he was stupid, that was okay. If they thought he was violent, that was okay. If necessary, Mozasu was not afraid to clean out all the teeth right from their mouths. You think I’m an animal, Mozasu thought, then I can be an animal and hurt you. Mozasu did not intend to be a good Korean. What was the point in that?
In the responses to Eileen’s and Kalos’ letters, I have seen many point out this is not only a generational divide, and that many in my generation hold differing views as well. You are correct. This is not a generational issue. It is not so simple as declaring that all our parents are racists and that we, the second generation educated elite, are above prejudices and here to save our parents from racist beliefs. It is preposterous, however, to deny that many Chinese Americans are racist towards Black Americans. Almost all of us have heard or even used the term 黑鬼 (hēi guǐ - black devil) as a slur. The simple existence of this term is evidence for the prevalence of racism within the Chinese American community.
The line between stereotypes and useful generalizations is a thin one, and there are times we have all crossed this line. But by engaging in more of the thoughtful dialogue Eileen’s letter has generated, we can all work together to dismantle our prejudices and promote anti-racist thinking in order to create the more equal, more judicial, and greater America we all call for.
When Obama declared in 2004, “there’s not a black America and white America and Latino America and Asian America; there’s the United States of America,” he claimed our shared American identity mattered more than our differences [1]. Yet, in light of George Floyd’s murder, it is clear just how much divides our country today. Perhaps it is better to see Obama’s words as a promise, another in a long line of promises made by great Americans like Abraham Lincoln and Martin Luther King Jr. Progress takes time, but as James Baldwin expressed over forty years ago [2]: “How much time do you want for your progress?”
Yitao, I too, do not condone the violence and damage done to small businesses as a result of these riots. As you say, these actions are “totally counter productive to solving the problems.” However, it is important to make a distinction between rioters and protestors. We can not let the actions of the rioters drown out the voices of the many more protestors who will not tolerate waiting anymore. The actions of rioters speak to the severity of the issues, but we must not let the debate over these rioters derail the necessity of making meaningful contributions toward police reform and anti-racism.
Best,
Matthew Shu (束骏杰)
[1] B. Obama, “Barack Obama’s Remarks to the Democratic National Convention,” The New York Times, Jul. 27, 2004.
[2] K. Thorsen, James Baldwin: How Much Time Do You Want For Your “Progress?”
[3] G. Kao and M. Tienda, “Optimism and Achievement: The Educational Performance of Immigrant Youth,” Social Science Quarterly, vol. 76, no. 1, pp. 1–19, 1995.
[4] Min Jin Lee, Pachinko. Grand Central Publishing, 2017.
[5] J. Lee and M. Zhou, The Asian American Achievement Paradox. Russell Sage Foundation, 2015.
[6] J. Yiu, “Calibrated Ambitions: Low Educational Ambition as a Form of Strategic Adaptation Among Chinese Youth in Spain,” International Migration Review, vol. 47, no. 3, pp. 573–611, 2013, doi: 10.1111/imre.12037.
]]>