Oisin Daly Kiær, Paul Grau, Yoo Jin Lim
Given a massive volume of vocabulary words out here, learning vocabulary through standard flashcards yields poor results, due to the individuals’ lapses in motivation and inspiration in the face of tedious work, along with a lack of good guidelines to select what and how to learn.
Learning vocabulary is a fundamental activity that affects a wide range of academic skills from writing to foreign language acquisition. Despite its unquestionable value, vocabulary learning is often considered the biggest pain point by many students as they progress through their school years. Many students find the precise activity of learning to be tedious and even frustrating. This in turn presents additional challenges on teachers to motivate their students to learn effectively on a regular basis.
Recent advances in mobile and ubiquitous computing have shown potential to alleviate these issues. Since the mid-1990s, research in Mobile-Assisted Language Learning has focussed on using mobile systems to help students, who have repeatedly shown high interest in these solutions. A number of projects have tried to employ findings in learning theory (e.g. spacing effect), motivational effects (e.g. gamification), and other fields of research with promising results, but few have been able to sustain engagement.
One author’s personal account: From my personal struggles to learn vocabulary as a student and later to impart the knowledge as a teacher, I have gained valuable insights into vocabulary learning with flashcards. I was determined to perfect my vocabulary skills for the SAT, so I picked up a list of 5000 vocabulary words and began my solitary journey through it with flashcards as my only learning tool. It was a painstaking process of trying different ways to create flashcards to learn the words, all the while dealing with the sheer volume of 5000 words on my own. In the end, I went through five different designs of flashcards to come up with one that made learning effective and fun. When I became a teacher, I faced the challenge of finding interesting examples or images to show with the vocabulary words. In retrospect, I realize how these issues could have been effectively tackled by crowdsourcing, specifically by a crowd of vocabulary learners.
A big part of learning is personal involvement. This is not achieved intrinsically with the expert-based approaches that currently exist (pre-made flashcard decks consumed as-is, studied alone). By employing a learnersourcing approach, where the crowd is brought together by their hunger for knowledge, and having them cooperate or compete with each other, we attempt to foster a sense of community that gets users more invested in the process. Creating and consuming at the same time, and making the affair a social one makes for significantly better learning. Machines will always be part of the solution, but for things like finding relevant sentences using vocabulary words, finding relevant pictures, or fixing a learner’s mistakes, machines are still insufficient because they cannot reliably judge the quality or learning effect.
How might we improve vocabulary learners’ ability to select which words to study?
How might we output flashcards with higher quality, for reuse elsewhere?
How might we involve users with each other to help them learn socially?
How might we exploit naturally occurring idle time for vocab learning?
How might we motivate people to study more regularly?
How might we make the vocab studying experience to be exciting and life-changing?
How might we make people feel more accomplished? help them experience firsthand how vocabulary improves their writing skills? Conversational skills?
How might we improve vocabulary without specifically studying a list of single words?
How might we help teachers teach vocabs more effectively?
How might we make the list of vocabs like a blockchain?
Flash Idol taps into the experience and creativity of vocabulary learners to seek out the best flashcard design
Scenario A:
Alice has taken the SAT twice, but her critical reading score does not seem to be improving. She knows that using flashcards helps with learning new vocabulary but is not sure what to write on them to optimize her learning. Her flashcard design is currently stuck at the simplest format of putting the vocabulary word in the front and the definition in the back. So, she goes on the Flash Idol site to check out how other learners make their flashcards. She is greeted by a gallery of creative flashcard designs, coming in different various arrangements of words and images. She decides to go with the No. 1 design and eagerly implements it onto her new flashcards.
Scenario B:
Bob is an overachieving student who spends all his spare time on learning vocabulary from his handmade flashcards. Over the years, he has experimented with various elements on his flashcard to come up with the design that best serves him. Unlike the conventional flashcard, his flashcard contains both the vocabulary word and definition on one side. He believes this helps to create a single mental image in his memory. He often gets new words to study from the Internet, and while he is at it, he decides to enter the Flash Idol competition with his flashcard design. A few days later, he is rewarded with the honor of being voted as the No.1 Flash Idol in the world and receives an interview request from a test prep book publisher.
Motivation - A crowd worker would do this for reputation and altruism. The ranking system and competitive setting offers a chance to be publicly recognized for their design. For others who do not win, they may participate for altruistic reasons, to share their knowledge and creativity.
Aggregation is done by collection. It is essentially collecting various design ideas from learners to seek out the best design.
Crowd pool is made up of voluntary workers who have studied or are currently studying vocabulary with flashcards. They may be students, parents, teachers, or anyone with interest in vocabulary improvement.
Quality control is done by the competition model. The prospect of becoming the Flash Idol is designed to elicit quality submissions from the participants.
Human Skill - the human skills required for design submission include basic English skills and minimal artistic instincts.
Process Order - Requester to Worker, since Flash Idol posts up the design challenge as the main requester and learners directly submit their designs without additional computation.
Goal Visibility - The overall goal of the system is clearly visible. The design goal and the potential learning benefit will be clearly described in the Flash Idol page.
FlashCrowd is a collaborative flashcard generation/studying application for English vocabulary, where users iteratively improve on a shared pool of cards as they study them for their own benefit.
Mwenye emigrated from Uganda to America at 13 with his mother. He went to high school in upstate NY, and was a high performer in math and science. His English is good, if accented, but he desperately needs to increase his vocabulary if he wants to score well on the SAT. His teacher recommends FlashCrowd for this purpose, so he tries it out. When Mwenye fires up the app, he sees a list of decks he can study, labeled according to the domains or specific tests they cover. He decides to explore a little, and taps the free-for-all deck. It contains ~17000 cards. He flips through the cards for a while, chuckling at the esoteric and often rude words on display. The cards contain their respective words, along with a definition. Almost all of them contain a picture or example sentence. Some cards contain multiple, similar definitions or examples, marked in red as being in conflict. There is an option to vote on which of the option one prefers. At some point, Mwenye finds a card for the word ‘manifold’. It was clearly written by someone who works with cars, because the definition and example picture only talk about an exhaust manifold. Mwenye knows the word from mathematics, so he decides to make a contribution. He presses the ‘edit’ button, and adds an alternative definition to the card. He is given the option to explain his edit. He writes: "The definition of this word is manifold :)", and submits. Mwenye decides it’s time for work, and he switches to the SAT deck. He spends time studying the deck each day, while on the train or toilet, or for dedicated study sessions. Being a completionist, he does not like it when a card is missing a feature (pronunciation audio clip, picture, example sentence etc.), so he spends some time finishing these cards when he finds them. As the time for the SAT approaches, he transitions into more structured practice, where the app puts his skills to the test. He aces the SAT, and lives happily ever after.
Motivation - The crowd workers are motivated by their personal desire to learn. They might use this platform over a regular pre-made flashcard deck because they appreciate the sense of community and influence we give.
Aggregation - Individual workers’ work is combined through an iterative improvement process of Fix-Verify. It seems extraneous to add a Find step to the start of this. New cards can be made too, of course. The community should verify whether they belong in a given deck or not.
Crowd pool - The crowd pool is people who know English, but want to increase their vocabulary or help others do so. This is one of the benefits of large standardized tests like the SAT or TOEFL, it creates a demand for learning of the skills that they test.
Quality control - The iterative writing loop of (Find-)Fix-Verify adds multi-level review mechanism, which should ensure that the quality of the cards increases with time.
Human skill - The participants just need to know English. The community can sculpt the selection of cards as it sees fit, so people of lesser or greater language skills can be catered to if there are enough of them. The system could be expanded to beginner level English or to other languages as well.
Process order - Computer → Worker: The Computer does the initial job of creating simple flashcards with sentences and definitions from online. The workers are also the requesters in the sense that they are fixing and looking for flashcards at the same time.
Goal visibility - The goal is intrinsic to the task and clearly visible. As you work, the flashcard decks become larger, and more knowledge is pooled for people to take advantage of.
VoCha (Vocabulary Chat) is an interactive platform that pairs up a small number of people to collaborate in real-time on producing high-quality example sentences, which provides a fun way for both learning new words’ usages as well as giving feedback and contributing to a collection of good study material.
Scenario A:
Seongmin is studying for the TOEFL and wants to improve his vocabulary. His Hagwon gave him a long list of vocabulary, but it is too boring to study that. He found a similar set of words on Memrise, but it is lacking good example sentences and he is not sure about the selection, so he gives VoCha a try.
After logging in, he is presented with the interactive interface which shows him a word along with one of the good-quality examples created in the past, and explains the interaction. When Seongmin is ready, he can join the real-time crowd. A small group of about 10 people works on coming up with a good example sentence for a new word. Seongmin has never heard that word before, but some others have, so he uses the chat to clarify the meaning for everyone. One person suggests a first example sentence, and together they work on improving it. The editor supports features like Google Docs for suggestions and deleting, and also a way to upvote or downvote others’ suggestions. Once everyone is happy with the example and all questions are cleared up, the users click on "Finish". The system now ranks each individual contribution. Seongmin didn’t change anything in the sentence because he didn’t feel comfortable with the new word, but his interaction in the chat and upvoting the others’ suggestions earned him his first 32 points. He also learned a new word and its usage, so he is eager to continue.
Scenario B:
Jieun is a professional writer. She also worked as a teacher for a while and enjoys helping others learn languages, so she joins the VoCha crowd. After participating for some time, her reputation score is pretty high.
The score is used to smartly match up people: If Jieun would always be in groups of only low-level students, she might lose motivation because she would be the only creative and authoritative source. Instead, VoCha pairs up 2-3 high-level users, 3-4 mid-level users and 1-2 entry-level users. This way, in every session there is a chance to learn something new, deepen existing knowledge and also enjoy the creative, collaborative process.
Jieun chose to make her profile public and link to it on her LinkedIn. A few weeks later she gets to join a news company employing crowdsourcing for their stories. The CEO says she was impressed by Jieun’s activity on VoCha which shows her language skills and engagement with the crowd. In her free time, Jieun continues to use VoCha as a fun way to interact with others.
Motivation - A) Fun way to learn with good material, B) Providing feedback and contributing to a crowdsourced library of high-quality learning study material can give the user a sense of accomplishment. For both, a reputation system helps for continued motivation.
Aggregation - Workers collaborate in real-time to find the best solution, by using a synchronised editor as well as a chat.
Crowd pool - Mostly language learners, but also people interested in helping in education
Quality control - The group works together to ensure good quality. They are incentivized by a reputation system. The system is used to smartly match both high-quality and low-quality workers together.
Human skill - No special skill is needed except for the willingness to learn.
Process order - The initial list of vocabulary is created automatically. Workers then collaborate to improve the list and add example sentences. These micro-tasks are finished in small steps while the whole collection improved continuously.
Goal visibility - Depending on the final design, either low (the goal is learning more and gaining reputation), or high if the list of vocabulary that is being worked on is finite (the user can be shown how many words they have studied already)