1 of 3

This material is based upon work supported by the National Science Foundation under Grant No. #1552114.

Understanding Linguistically Diverse Students’ Development of Vocabulary During Science Learning Using Semantic Network Analysis

Yinuo Hu

Computer Science B.S. & Sociology

Faculty Advisor: Kihyun "Kelly" Ryoo, Ph.D., School of Education

2 of 3

Introduction

  • Engaging linguistically diverse students in developing, revising, and explaining scientific models in pairs can promote their understanding of unobservable scientific phenomena through discourse-rich practices.
  • Exploring linguistically diverse students’ conservations may reveal patterns of their vocabulary development during scientific modeling practices.
  • Previous work showed that Semantic Network Analysis (SNA) has the potential to visualize the structure and relationships among the use of vocabulary.
  • Given the limited research in applying SNA on pairs’ discourse during scientific modeling practices, the study explored the following research questions (RQs):

Methods

  • As part of a larger NSF project that explores how visualizations can improve 8th-grade linguistically diverse students’ science learning, students worked in pairs to develop models and write explanations of how thermal energy affects the state of water molecules during a phase change.

RQ1: What’s the structure of linguistically diverse students’ vocabulary and the relationships among the words?

RQ2: How does students’ vocabulary development change over time?

Semantic Network Analysis (SNA)

  • Data was cleaned up by removing stop words (e.g., the, a) and lemmatization.

​

  • Any words that were used more than three times were included in a word list.
  • Building on the existing list from the larger project, this SURF project added additional words from the analysis to revise the Three Tier Model (Beck et al., 2002) through iterative processes.

changes, changed, changing

change

Methods

Divisions of Modeling Activities

  • Students’ talk turns were divided into six divisions based on the following:

Building and revising models

Writing and revising explanations

D1

D2

Receiving the 1st feedback

Working on models

D3

D4

D5

D6

Receiving the 1st feedback

Working on explanations

Half point of model revisions

Half point of explanation revisions

Tier 3: Specific Content Words

e.g., water molecule, state

Tier 2: General Academic

Words

e.g., describe, explain

Tier 1: Basic Words

e.g., ice, water

  • This SURF project used 16 transcripts from the video files that were collected and analyzed as part of the larger NSF project.
  • This resulted in 5649 talk turns of pairs.
  • Pairs used automated feedback to revise models and explanations.
  • The project used KBDeX, R, and Gephi to visualize the semantic networks.

3 of 3

Results

Conclusion

  • With SNA, the study explored the structure and development of students’ discussions as they gradually centralized around several key scientific concepts.
  • Students made stronger connections of academic and content-specific words through discourse-rich collaborative modeling practices.
  • SNA can be more widely used in the future to understand the vocabulary development of students in different collaboration settings or in other disciplines.

D3

D1

  • Tiers 1, 2, and 3 words were mixed in D1.

D6

  • Some Tier 3 words (e.g., water molecule, thermal energy) were at the center of the network in D6.
  • Tiers 2 and 3 words moved closer and formed stronger connections among each other.

Tier 1

Tier 2

Tier 3

  • A cluster of Tiers 2 and 3 words showed better connections of key concepts.
  • Tiers 2 and 3 words were still not the center of discussions yet.

Zoom-in figures of D1, D3, and D6

  • Basic words were the center of the D1 network.
  • Tiers 2 and 3 words were weakly connected.