Word Recognition

Platform: HTML · German · verified 89d ago

This is a task in the LEAPS Zurich study. Click to know more.

Local data export available

This task saves data locally in your browser and offers a download at the end of the session. For production use, you will need your own server-side persistence.

Description

Pixi.js + TypeScript educational game built for the LEAPS Zurich learning-progress study. The child hears a spoken target (a real German word like "Rad" or a pronounceable pseudo-word) and is shown four candidate spellings; they tap the one that matches the audio. The task is split into two parallel arms — real German words and pseudo-words — each with its own demo, practice and test scenes. Real-word items probe sight-word recognition; pseudo-word items probe phonological decoding by removing the option to retrieve a known lexical entry. The flow is Demo → Practice (real) → Test (real, 31 items) → Practice (pseudo) → Test (pseudo, 31 items). The test scenes cap at 7 minutes each and show a 'stop frog' early-exit overlay after 30 seconds of inactivity. Each trial logs the four candidate spellings, every press/release event, every audio replay, the final selection, and per-trial timestamps.

Contact & Lab

Lab

Developmental Cognitive Neuroscience Team

kjpd.uzh.ch

Principal investigator

Silvia Brem

silvia.brem@kjpd.uzh.ch

University of Zurich

Task contact

Iliana.Karapidis@kjpd.uzh.ch

Contributors

Iliana Karapidis

University of Zurich

Conceptualization

Supervision

Nicoletta Klotz

University of Zurich

Data Curation

Investigation

Hugo Farajallah

University of Geneva

Software

Citation

No citation available — contact the task author.

Credits

No credits provided.

Domains

Vocabulary

Reading

Psycholinguistics

Part of suite:

LEAPS Zurich

Target Population

Age Categories

primary

Clinical Status

non-clinical

Additional Notes

German-speaking primary-school children, ages 6-8 (LEAPS Zurich cohort).

Languages

German

Run under

Engine

HTML / JavaScript

Format

Online + in-lab

Modalities

behavioral

Task Outputs

What it measures

Audio-to-orthography mapping in early readers, separating sight-word recognition (real-word block) from phonological decoding (pseudo-word block).

Output formats

json

One AppResults JSON per session. In browser preview, the file downloads automatically at the end. In LEAPS Android deployment, the same payload is forwarded via window.System.sendData to the LEAPS backend. Top-level shape: { practice: SceneResult|StopFrogResult|null, test: SceneResult|StopFrogResult|null, practicePseudo: SceneResult|StopFrogResult|null, testPseudo: SceneResult|StopFrogResult|null }.

Experimental design

Trial-based

62 trials

Within-subject: scene, word_type

Conditions

scene

demo

practice

test

— Demo (1 animated story item, no scoring), practice (2 items with feedback), test (31 scored items, 7-min cap). Each level runs once per word_type.

word_type

real

pseudo

— Real German words probe sight-word recognition; pseudo-words (pronounceable non-words) probe phonological decoding by blocking lexical retrieval.

Key measures for analysis

Real-word test accuracy (% correct on the 31 test items)
Pseudo-word test accuracy (% correct on the 31 testPseudo items)
Real - pseudo accuracy gap (lexicality benefit)
Mean response time per trial, separately for real vs pseudo
Audio replays per trial (proxy for decoding effort)
Selection-revision count (selections.length per trial)
Stop-frog rate (proportion of sessions terminated early)

Recorded variables

Variable	Role	Type	Description	Unit
`practice`	dependent	object	SceneResult for the real-word practice phase, or a StopFrogResult if exited early. Null if skipped.	-
`test`	dependent	object	SceneResult for the real-word test phase, or a StopFrogResult if exited early. Null if skipped.	-
`practicePseudo`	dependent	object	SceneResult for the pseudo-word practice phase. Same shape as `practice`.	-
`testPseudo`	dependent	object	SceneResult for the pseudo-word test phase. Same shape as `test`.	-
`answers`	dependent	array	Inside each SceneResult: an AnswerResult per trial.	-
`totalCorrect`	dependent	number	Inside each SceneResult: count of correct answers in that scene.	-
`totalTime`	metadata	number	Inside each SceneResult: total time spent in the scene.	ms
`reference`	independent	string	AnswerResult field: filename of the spoken target audio (e.g. wr_rad.mp3 for the word 'Rad').	-
`trialIndex`	metadata	number	AnswerResult field: 0-based trial index within the scene.	-
`startTimestamp`	metadata	number	AnswerResult field: timestamp when the trial started.	ms
`endTimestamp`	metadata	number	AnswerResult field: timestamp when the green-arrow submit was clicked.	ms
`answerOptions`	independent	array	AnswerResult field: the four candidate written words, each with isTarget (true for the correct option) and choiceIndex (1=top-left, 2=top-right, 3=bottom-left, 4=bottom-right).	-
`audioPlaybacks`	metadata	array	AnswerResult field: every time the spoken target was played, with timestamp and audioFile. Children can replay; the count is informative.	-
`selections`	dependent	array	AnswerResult field: every click on an option, with timestamp, choiceIndex, and isFinal (true for the one submitted).	-
`isCorrect`	dependent	boolean	AnswerResult field: whether the final selection matched the spoken target.	-
`responseTime`	dependent	number	AnswerResult field: time from trial start to final-submit click.	ms