Nexus Analysis for AI Literacy Research
Overview
Nexus Analysis, developed by Scollon and Scollon (2004), provides a methodological framework for examining AI-mediated literacy practices. Combined with Wertsch's (1991) Mediational Means concept, this approach treats digital interaction logs not as static transcripts but as time-stamped traces of social action.
Core Constructs of Nexus Analysis
1. Historical Body
The accumulation of prior experiences, habits, and dispositions that participants bring to any social action.
In AI literacy research:
- Prior educational history (years of prohibition-oriented schooling)
- Previous experience with AI tools
- Professional identity formation
- Source curation patterns reflecting "grounding truths"
2. Interaction Order
The patterns of social arrangement and roles that govern how people interact in specific situations.
In AI literacy research:
- Turn-taking patterns between human and AI
- Prompt constraint and evolution
- Latency as "chronometric signal of deliberation"
- Recursive prompting behaviors
3. Discourses in Place
The broader cultural, institutional, and technological narratives circulating in the research setting.
In AI literacy research:
- Institutional AI policies (or lack thereof)
- Professional ethics surrounding authorship
- Cultural narratives about "cheating" and assistance
- Tensions between efficiency and authenticity
The SPOC Model
A multi-dimensional analytical framework operationalizing Nexus Analysis for AI interaction study:
| SPOC Dimension | Focus Question | Critical AI Literacy Link | Observable Behavior |
|---|---|---|---|
| Source Selection | Did the student bound the AI's knowledge base? | Transparency & Accountability | Uploaded relevant domain-specific texts; references sources in prompts |
| Prompting Depth | Did the student use AI for cognitive enhancement? | Continuous Learning & Innovation | Iterative refinement; layered contextual prompts; professional personas |
| Output Evaluation | Did the student demonstrate skepticism? | Exploration & Evaluation | Follow-up prompts for verification; bias detection; cross-referencing |
| Critical Integration | Did final work maintain human voice? | Human-Centered Approach & Agency | Substantial revision; personal examples; rejection of generic content |
Digital Trace Data Analysis
Key Methodological Traces
-
Latency
- The temporal gap between AI output and student intervention
- Not "idling" but a chronometric signal of deliberation, critical reading, or repair labor
- The "physical trace of cognitive friction in action"
-
Prompt Evolution
- Longitudinal shift in prompt architecture and constraint
- Reveals "trajectory of agency" - whether students mature toward orchestrated collaboration or default to algorithmic passivity
-
Edit Distance
- Quantified divergence between raw AI output and final artifact
- Forensic measure of "interpretive control"
- Captures the "labor of re-authoring"
Mapping Data to Theoretical Constructs
| Coding Component | Nexus Construct | Analytic Focus |
|---|---|---|
| Inputs | Historical Body | Prior literacy habits, source curation, professional background |
| Prompts | Interaction Order | Constraint-setting, turn-taking, epistemic stance assertion |
| Outputs | Discourses in Place | Tool constraints, citation accuracy, "AI voice" |
| Integration | Boundary Work | Human transformation, rejection, synthetic re-authoring |
| Reflection | Societal Discourses | Ethical justification, legitimacy concerns, professional identity |
Methodological Principles
- Process over Product: Analysis focuses on interactional traces rather than final artifacts alone
- Visible Orientations: Evidence must be interactionally visible, not inferred
- Digital Nexus Analysis: Acknowledging analysis of mediated interaction rather than physical observation
- Triangulation: Combining JSON logs with ethnographic reflections and participant annotations
Practical Applications
For Researchers
- Use time-stamped interaction logs (JSON format) to capture granular turn-by-turn evidence
- Code for pauses, revisions, and source-grounding behaviors
- Treat latency combined with high edit distance as "smoking gun" for critical AI literacy
For Instructional Designers
- Design learning environments that make the "evaluative loop" visible
- Create structured reflection prompts that surface boundary work
- Build in pause-and-probe moments for deliberate AI engagement
Key Sources
- Scollon, R., & Scollon, S. W. (2004). Nexus of Practice
- Wertsch, J. V. (1991). Mediational Means
- Mollick, E., & Mollick, L. (2023). Practical AI use in education