1 of 14

����

What do the most popular �language learning apps teach? �The tension between programming ease �and acquisitional needs

Thor Sawin & Gabriel Guillén

Middlebury Institute of International Studies

tsawin@miis.edu & gaguillen@miis.edu

03/20�#AAAL2017

2 of 14

What the average app downloader experiences:

Low-Hanging Fruit?

Problem > Literature > Methods > Results > Discussion

3 of 14

“For the most part, apps are not vehicles for new ideas; instead, they usually digitize old ones (...)�The app (...) throws some pop-psychology jargon at those who visit its website (“adaptive learning algorithm,” “confidence-based repetition”), but you don’t need to have read Wilhelm Wundt, or even Malcolm Gladwell, to understand what the app does. It’s a platform for digital flashcards” (New York Times, 2014)

Problem > Literature > Methods > Results > Discussion

4 of 14

Problem > Literature > Methods > Results > Discussion

Nation

2001

Learning a word is more than learning isolated units of language

Cruz-Piñol

2002

Only 4 out of 207 Spanish sites allowed oral production interaction

Kukulska-Hulme & Shield

2008

Lack of social contact and collaborative learning in MALL (Mobile-Assisted Language Learning)

Chapelle (2001, 2009) Hubbard (2006) CALICO Review Guidelines… Pegrum (2014) 4 Types of MALL

5 of 14

↓MALLs

  1. Content (input)
    • Positive evidence of deployment of structures by human authors
  2. Creation (output)
    • Creating discourse for other human users to process
  3. Communication (interaction)
    • Exchanging information with another language user
  4. Tutorial (automated feedback)
    • Negative/positive evidence about language deployments by learners

  1. Phonology-to-orthography
    • <defend> ←→ [di’fεnd]
  2. Lexical semantics
    • L1 translation, TL synonyms, picture illustrating core sense
  3. Lexical grammar (verb-argument construction)
    • DEFEND s.o. FROM s.t.
    • DEFEND one’s {thesis, honor, reputation}
    • {valiantly, forced to, bravely} DEFEND
  4. Functional morphology
    • Defended, defending, defender, defense, defendant
  5. Sentential syntax
    • [Subj [[Verb [NP]] NP] or [Det [Noun [CP [who] [Sent]]]
  6. Sentential semantics
    • Trump says US must be paid more to defend allies.
  7. Pragmatics & Linguaculture
    • Puedes ponerme un cafe vs. I’ll do a coffee

Problem > Literature > Methods > Results > Discussion

Ling. Systems →

6 of 14

Info on # Downloads unavailable

Hit algorithms includes: (un)installs + reviews + engagement + recency

�Which apps show up most in 6 “Top 50” searches?

  1. iPhone Store “Language learning”
  2. iPhone “Learn language”
  3. iTunes “Language learning”
  4. iTunes “Learn language”
  5. Android “Language learning”
  6. Android “Learn language”

Problem > Literature > Methods > Results > Discussion

7 of 14

Clear cut off at 48 apps, represent 37 software designs(Busuu Spanish vs. Busuu) (Mondly: English vs. Mondly German)

37 app designs:

Rated by two raters for which of 4 “MALLs”

  • Primary (1) = Unavoidable affordance
  • Secondary (0.5) = Optional affordance
  • None (0) = No affordance

Rated by three expert raters for which of 7 linguistic subsystems

  • Explicit Primary (3) = Unavoidable explicit teaching
  • Explicit Secondary (2) = Optional explicit teaching
  • Implicit (1) = Opportunity for implicit learning
  • None (0) = No opportunity for even implicit learning

Problem > Literature > Methods > Results > Discussion

8 of 14

Problem > Literature > Methods > Results > Discussion

Primary Focus

Secondary Focus

No affordance

2 rater weighted kappa = 0.59

9 of 14

Problem > Literature > Methods > Results > Discussion

Explicit Primary

Explicit Secondary

Incidental possible

Incidental impossible

3 rater weighted kappa (mean of means) = 0.47

10 of 14

Problem > Literature > Methods > Results > Discussion

App name

Acquisitional Function

Linguistic Focus

“Typical” Apps

Memrise, Lingvist, Nemo, StudyCat, Quizlet

  • Tutorial primary focus
  • Content secondary focus
  • Phonology/orthography
  • Lexical semantics

Atypical Apps

HelloTalk, Tandem

  • Communication primary
  • Creation secondary
  • Incidental focus on all skills, Explicit on none

Busuu

  • Tutorial and content primary
  • Creation and communication secondary
  • Explicit primary focus on both vocabulary and sentential syntax
  • Explicit secondary on morphology and pragmatics

11 of 14

  • Can we have an app with all 4 MALLS and 7 subsystems?
  • More “attractive” for beginners or casual users (exceptions)
  • Most popular apps are not necessarily the “best” ones
    • Learners at the mercy of marketing claims and SEO
  • Need of connecting academic rating (CALICO) with learners
  • Still in the “flashcard era” of MALL…
    • Digital flashcards with recordings… isolated recordings (still treating words as isolated units of language)

Problem > Literature > Methods > Results > Discussion

12 of 14

  • Form-Meaning-Use: Lexical grammar + pragmatics; Communication + creation
    • Less than incidental focus on use
  • Lexical semantics + phonology; Tutorial
    • Boring, useless, but cheap enough to turn a profit
  • Learners already believe language = words
    • Fail at apps → bad language learner, impossible feat
  • Well-balanced app - higher engagement
    • but more expensive,
    • requires critical mass of users

Problem > Literature > Methods > Results > Discussion

13 of 14

App Diet

One Stop App

Problem > Literature > Methods > Results > Discussion

14 of 14

Top 37 Apps

Babbel

DuoLingo

Memrise

Busuu

RosettaStone

Google translate

Learn 50 languages

Mango languages

LingQ

Tourist language

Bravelol

Learnlanguages

ChineseSkill

Innovative Language 101

JW language

learn & play

Hello chinese

Learn German - (ye li)

Learn japanese easily

Awabe

lingo arcade

Lerni

Lingvist

Nemo: Spanish

SpeakTribe Spanish

uTalk

HelloTalk

Mondly

Quizlet

Tandem

6000 words - learn spanish

Beelinguapp

Daily Dose of Language

Drops (learn japanese)

FluentU

Studycat (fun Spanish)

Human Japanese lite

In 24 hours