LT5461 in-class experiments

Browser experiments for LT5461 Cognition and Language Differences (City University of Hong Kong). Students run them in class. Their anonymous data go into a Google Sheet, and the class results update live on the projector.

• Students: https://psycholinguistic-experiments.github.io/experiments/ (lists all experiments)
• Projector: https://psycholinguistic-experiments.github.io/experiments/results.html
• Code: https://github.com/psycholinguistic-experiments/experiments (GitHub Pages, main branch)

Page	For	What it is
index.html	students	Start page listing all experiments
word-task-1.html	students	Masked translation priming (60 ms prime), lexical decision
word-task-2.html	students	Visible translation priming (200 ms prime), lexical decision
shape-task.html	students	Sound symbolism (bouba/kiki), 3 × 2
judgement-task.html	students	Foreign-language effect: gambles, sunk cost, superstition (Chinese vs English)
association-task.html	students	Association task (FAST): chained associations in Chinese and in English
sentence-task.html	students	Sentence verification: the typicality effect (true/false statements)
bird-task.html	students	Bird task: how typical 12 pictures are as birds (1–5), mapped as a semantic space
report-task.html	students	Report task: crime framed as a beast or a virus (Thibodeau & Boroditsky, 2011), in Chinese or English
animal-task.html	students	Animal task: arrange 20 mammals by similarity, then 30 similarity-based inferences (after Douven et al., 2023)
fluency-task.html	students	Fluency task (in Chinese): two 120-s semantic fluency blocks (animals; fruits and vegetables), then joyous exploration and deprivation sensitivity from the 5DC
reading-task.html	students	Reading task: self-paced reading of 30 passages with or without “because”, a recognition probe and some comprehension questions (after Millis & Just, 1994)
colour-task.html	students	Colour task (Mandarin): name 14 blue-green colours twice, then 60 colour-matching trials; are pairs with different names matched faster?
emotion-task.html	students	Emotion task (in Chinese): arrange 47 emotion words by meaning, as in Chaouch-Orozco, Li & Li (2026)
results.html	you	Live class results for the projector
apps-script/Code.gs	you	The Google Apps Script behind the data

The site is plain HTML, CSS and JavaScript. There is no build step and no dependencies.

In class, every year

1. Before class, rehearse by adding ?test to any task address, for example …/word-task-1.html?test. Test runs are filed under a separate test-<date> class, so they never mix with the real data.
2. Put the start page on a slide, as a link or QR code.
3. Open results.html on the projector.
   • It shows today's class and updates every 5 seconds.
   • Hide results shows only the running counts while students are still working. Untick it to reveal the results.
   • The Class menu shows earlier dates. All 2026 classes pools the whole year, which is useful for comparing years.
   • Download this class (CSV) gives one row per student.
   • Each tab opens with a short Summary: one line per finding, with its test and whether it is significant. It stays hidden while Hide results is ticked.

Nothing needs changing from one year to the next. Each class is simply the Hong Kong date on which the students did the tasks.

The data

All data go into the Google Sheet (personal Gmail) that the Apps Script is attached to. There are two tabs per task (masked_*, visible_*, bouba_*, fle_*, fast_*, svt_*, birds_*, report_*, animals_*, fluency_*, reading_*, reading_start_*, colour_*, colour_start_*, emotion_*; the report task sends no trials, so it has only report_people):

• *_people: one row per student, with the numbers the results page uses.
  • Word tasks: priming effect, condition means, accuracy, screen refresh rate, and the measured prime duration.
  • Masked task: also the awareness answer.
  • Shape task: the six cell means.
  • Judgement task: language, the three scores, the checks.
  • Association task: group, language order, lists, the language questions, the ten 0–10 self-ratings, and every answer (each chain as two fields, c_<seed>_a = language + answers 1–5 and c_<seed>_b = answers 6–10). No valence is stored: the results page scores the answers each time.
  • Sentence task: median RT for high- and low-typicality true statements, the typicality effect, accuracy, valid-trial counts, time-outs.
  • Bird task: one column per bird (house_sparrow … flying_fox, 1–5), the median rating time and the five English self-ratings.
• *_trials: one row per trial. Every trial's forward-mask and prime durations are measured and stored.

Students are identified only by a random code (e.g. K7Q-3FD) kept on their device, and the same code links one student's tasks. No names, student numbers, e-mail or IP addresses are collected. These are teaching demonstrations. Using the data for research would need ethics approval and a consent procedure first.

Google's web-app endpoint occasionally answers with an error page (about 1 request in 8 when several arrive together), so every request retries automatically. A submission that still fails waits on the student's device. It is retried every 20 seconds while the page stays open, and again the next time any task is opened. A repeated submission is never stored twice.

This is the beginning of the README. Read the rest at the original. Read the full README ↗