The Conundrum of Vocabulary Games

The Conundrum of Vocabulary Games

Because I should be grading research papers I decided to fool around with the design constraints of vocabulary games for a friend curious about adding some games to his graded reader app. 

QUESTION: How hard is it to develop good* vocabulary games that draw on the vocab from a text.

*good = not a multiple choice quiz.

TL;DR: building something new and interesting is possible, but not easy. Here's why:

A short primer on vocabulary games before we get started:

The simple reason that there aren't many good vocabulary games is that the constraints are pretty high. 

The prime constraint is the first tenet of games: "games are a series of interesting choices" (Sid Meier). This basically means that good games invest the player with a sense of interest or engagement through the choices the game presents. In low-narrative games, like match three-games (Candy Crush), or traditional board games (Checkers, etc.) or something like Angry Birds, the game presents multiple binary choices (do this now or do that now, but not both); the interest comes from deciding which choice to make, and in the case of dexterity games, performing that activity correctly. So in Angry Birds the decision is "which part of the structure am I most likely to be able to knock down given the bird-type I'm firing and the difficulty of hitting that part of the structure?" That’s an interesting choice. Put that in your game and ba-pa-bippidy-doo, you’re pushing plush dolls and action figures and shitty bird-themed movies. 

An example of incorporating interesting choices in a vocabulary game is a board game that I use in class. It uses the mechanics of Reversi (Othello), the game in which you capture the opponent’s pieces by bookending them with your own, thus flipping them to your color. In the classroom game there are two teams, with a set of cards that function like the black and white sides of reversi pieces (blue on one side, yellow on the other), but also have text written on them. A target word is written on one side of each card and a definition or synonym is written on the other. The choice that the players make is “which move is the best to capture board space (or deny an opportunity to my opponent) based on my chance of correctly guessing the information that is on the other side of the cards I intend to flip.” So for example, a player may have the choice between trying to flip six cards to their color, four of which she is unsure of answering correctly, or flipping three cards that she’s certain to get right, with each choice presenting a different opportunity cost (each choice influences the progression of the game directly, by giving or denying future opportunities). It’s pretty engaging, especially when played in teams, and incorporates a good deal of review as the cards in the middle of the board get flipped back and forth repeatedly. 


The Reversi game could be adapted for the digital sphere, but would require 1) some simple AI, and 2) multiple-choice options for selecting answers (rather than saying what’s on the other side and then looking at the card, as learners do in class). The multiple-choice element would lessen the engagement, however, as it would forefront the quizziness of it. As with most competitive games, playing against another person is more engaging than playing against the computer, as well, so a single-player version wouldn’t have the tension and unpredictability of a live opponent. 

We could spend all day talking about different kinds of choice in games, but since we’re by necessity limited to low-narrative games, so we’ll just stick with those for now (there are plenty of potentially great narrative games, but, sadly, they need predetermined vocabulary sets, large budgets, player investment, and don’t look good to admin types). Here’s a link to a mock-up of a reading fluency game I designed when the last ELL website CEO (Arnar Jensson at Cooori) asked me to design a reading/vocab game (the player is just typing in the cloze blanks and selecting the characters) ( The primary issue with the design is the difficulty of matching it to specific curriculum or outcomes. That is, it would make a decent direct-to-end-users game, but isn't something that teachers could slot into their curriculum easily (also, not a vocabulary game, so back to the point).

In your typical quiz “game,” the choice (e.g. which of the four possible answers is correct), is simple and not engaging because there is isn’t really an alternative choice; it’s just a choice between correct or incorrect, and if you know the correct one, there isn’t even a decision to be made. In addition, the choices are discrete -they don’t impact the next event in the game, as, for example, removing a set of five red gems in a match-three game reshapes the playing field with positive or negative outcomes, or the above example of Reversi.  

One way that quiz games are made more interesting is by employing additional mechanics (mechanic = game rule or condition) that apply pressure or cognitive load to the choice: 

·      Players have a limited amount of time to answer each question

·      Players have to answer as many questions as possible in 10 seconds (which does present a somewhat interesting choice of “do I spend time figuring this out or skip it?”)

·      Players are playing against another person (adds an element of “perceptually authentic” randomness)

A second way that quiz games are made more interesting is by the introduction of an element of randomness or unpredictability, such as drawing from a large question bank that has multiple topics, peculiar questions, or questions of varying levels of difficulty (Trivial Pursuit).

A third way to increase engagement is through finely-tuned, appealing UI. You can see this in tools like the Duolingo app or in the gorgeous-but-fatally-flawed Spell Up game from Google. It can’t MAKE a game or activity, but appealing UI can persuade the user to engage with rudimentary activities. Unfortunately, the reverse is true: bad UI can break anything.

The elements above can combine to create some really great quiz games, such as QuizUp, a very popular 1) competitive (multiplayer only) trivia quiz game that features 2) a large question bank and 3) a timer, all wrapped up in 4) a sweet UI.  

A common design mistake in educational games is the chocolate-covered-broccoli approach. That is, there is some element of the game (the chocolate) that is intended to make the learning part (the broccoli) more appealing. So for example, a few years back there was a game where players were an Indiana Jones character navigating a temple (chocolate) and to succeed had to answer vocabulary questions (broccoli). The intention was that players would answer the questions in order to “enjoy” playing the rest of the game, but navigating the temple and answering the questions weren’t thematically or mechanically linked. These games are generally quite bad, because players need to be engaged by the current activity, not the hope of a future one. There are a few apps like Cambridge’s Phrasalstein and Phrasal Verbs Machine that pull this off, however (they also aren’t games). 

So, to recap, the problem with word-and-definition-type vocabulary “games” (quotes because they are actually gamified activities, not games) is that there are relatively few design choices available, and most of them have been used, reused, and used poorly to the point of tedium. That’s why (I believe) designers for the top language-learning tools, like Duolingo, don’t do “games” at all, but rather focus on enriching the experience of activities through attention to UI and pacing. 

Let’s take a look at the constraints of a vocabulary game/activity based on a set of graded readers and see what we could do. I’m assuming the following are fixed constraints:

1. Single-player (multiplayer would unleash a torrent of awesome, but I’m gonna assume that’s too gamey/unpredictable for most education system-facing applications).

2. Browser-based, with limited HTML-5 graphics and animations, mobile optimized

3. Locked down (no integration with SNSes, etc. -players aren't going to be messaging each other or a chatbot)

4. Short games

5. Beginner to advanced levels

            6. Metrics (scores recorded and reportable)

OK, the (somewhat) flexible constraints:

C1: Unknown / unlimited item set

It’s nearly impossible to work with an unknown data set, such as if the players can use any words from the readers in a game. Doing this would pretty much limit us to quizzes, primarily because the larger the range of options, and the more outfits you have, the more generic the game must become. Not good. One suggestion would be to have a set of generic flashcards/quizzes for all of the terms (cheap and easy!), and a game for a limited number of them. The flashcards and games would share the same backend.

C1 Question: Can we limit the number of playable items? Options: 

·      high-frequency words only (still a bit problematic)

·      game-compatible items only (high frequency AND easy to incorporate)

·      limited POS (no prepositions or phrasals, etc.)

C2: Definitions

If defining is the central theme of the game, we have a number of constraints:

1. Text-based or graphics intensive

The game is going to be text-based, which lessens player engagement (by increasing the cognitive load of the non-choice elements of the game), OR require graphics/animations for each item (matching pics to words, etc.), which would be great but increases costs exponentially. 

2. Database design

The way that the definitions are stored is a constraint here. Each item -let’s say the term “play”- would need to have a separate entry for each sense (3-4 entries for play as a verb, 2 for the noun form, for example). Depending on the api we're using, the definitions may not be available in this form (there may be a single text string with all of the definitions of “play”), meaning that we'd have to build it from scratch. With a limited number of items building the data base can be done manually and outsourced for a reasonable amount, but we need to know the structure of the data. 

C2 Question 1: Can we do non-definition related games? Options:

·      Spelling

·      Listening

·      POS (a match-three using parts of speech, for example)

·      Word forms

·      Grammar

·      Collocations

·      Contextual use (such as answering questions that use the word, or reading passages with the word)

C2 Question 2: Will we have a robust backend? 

C2 Question 3: Will we be able to pull sentences out of the readers? For example, can the game/activity extract five example sentences using the word “expand” from the graded readers?

C3:  Academic Tone

Lots can be done to make games engaging through humor, horror, quirky visual styles and the like. Unfortunately, for learners and administrators both (more so for admins), learning looks, well, boring. It looks like quizzes that ask serious questions and shame and pain for failure. It looks like endurance, not engagement. It is Word Engine

C3 Question 1: On a scale of 1-10, with 1 being laugh-out-loud silly, and 10 being “smells like learning,” what’s the rank for these games?

Wow, so that should get us started. As I said, I’m not overly optimistic about making a great compelling game given the constraints, but it’s worth thinking about. Even though generic quiz activities in a polished UI and a sweet UX are probably the way to go, I’ll keep dubbing around (with both games and activities) if you can let me know some answers to the above questions.


One Night Ultimate Werewolf

One Night Ultimate Werewolf

Whispers game

Whispers game