Progressive Sentences

Learning secondary language (L2) sentences and their meaning (e.g., English, L1) using a progressive presentation technique.

Publications

• Progressive Sentences: Combining the Benefits of Word and Sentence Learning, MobileHCI'25 Adjunct
  • Camera Ready PDF

@inproceedings{janaka_progressive_2025,
    author = {Janaka, Nuwan and Zhao, Shengdong and Ram, Ashwin and Sun, Ruoxin and Tan Jing Wen, Sherisse and Li, Danae and Hsu, David},
    title = {{Progressive} {Sentences}: {Combining} the {Benefits} of {Word} and {Sentence} {Learning},
    year = {2025},
    month = sep,
    doi = {10.1145/3737821.3749564},
    booktitle = {Adjunct Proceedings of the 27th International Conference on Mobile Human-Computer Interaction},
    series = {MobileHCI '25 Adjunct},
}

Contact person

• Nuwan Janaka (In)

Project links

• (private) Project folder: here
• Documentation: here
• Version info

Requirements

• Make sure Python3 (>=3.10) is installed ( e.g., Miniconda )
• Make sure npm and node is installed
  • Tested with Node.js v20.* (LTS)

Installation

• Install required libraries

  • wuggy using pip install wuggy
  • pandas using pip install pandas
  • OpenAI TTS using pip install openai
  • fastapi using pip install fastapi uvicorn
  • pygit2 using pip install pygit2

• [Optional] Install the following for the analysis

  • scipy using pip install scipy
  • statsmodels using pip install statsmodels
  • pingouin using pip install pingouin
  • seaborn using pip install seaborn
  • matplotlib using pip install matplotlib

• [Optional] Create the required credential files inside credential folder (if you want to use OpenAI audio generation)

  • Create a file credential/openai_credential.json with OpenAI credentials such as {"openai_api_key": "KEY"}

• Install the libraries for the frontend

  • cd frontend
  • npm install
  • npm run format

Application Execution

• Running the sentence learning

  • Run by python main.py
    • Then open the hosted URL (e.g., http://localhost:5173/) in a browser

• Generating L2 words

  • Run the L2 word generation via python generate_words.py after adding text/L1-words.csv
    • Results will be in output/L1-L2-words.csv
    • If there are issues with numpy, downgrade it using pip install numpy==1.*

• Generating L2 sentences

  • Run the L2 sentence generation via python generate_sentences.py after adding text/L1-sentences.csv and text/L1-L2-mapping.csv
    • Results will be in output/L1-L2-sentences.csv

• Generating audio (for L2/L1 words/sentences)

  • Run the L2 audio generation via python generate_audio.py after adding text/L2-sentences.csv
    • Results will be in output/audio/ directory

• Analyze results

  • Run the marking via python mark_answers.py after adding user_data/participants_answers.csv
    • If the script can not detect the correct one, you need to add marks by inputting correct values by commas
    • Results will be in output/participants_answers_marks.csv or output/participants_marks_summary.csv
  • To analyze marks, run python analyze_marks.py (assuming marks are at output/participants_marks_summary.csv)
  • To analyze ratings, use python analyze_ratings.py after adding user_data/participants_ratings.csv
  • To convert the long format data to wide format (e.g., JASP), use python convert_csv_to_anova_wide.py

• Generate L1 sentences from groups

  • Update the generate_mix_sentences.py with correct groups, and the nouns/verbs/adjectives based on groups
  • Generate the sentences using python generate_mix_sentences.py
  • To verifythat the generated sentences have a mix of groups,
    • Copy the sentences to text/L1-sentences.csv
    • Generate L2 sentences using python generate_sentences.py
    • Copy the generate L2 sentences to text/L2-sentences.csv
    • Run python analyze_sentences.py to get the results

• Verify all multimedia files (images, audio) are available

  • Run the analysis via python analyze_multimedia_files.py ,which verifies all the mentioned multimedia files in backend/data/Sentence_elements.csv are available

References

• Vue.js
• OpenAI Text to speech