Teacher Feedback, Student Self-Talk & Self-Concept: Detailed Study Notes

Article & Publication Details

• Full title: “The Impact of Teacher Feedback on Student Self-talk and Self-concept in Reading and Mathematics.”
• Author: Paul C. Burnett – Charles Sturt University, Australia.
• Source: Journal of Classroom Interaction, Vol. 38 (1), Spring 2003, pp. 11-16.
• Research setting: Six rural elementary schools in New South Wales, Australia.

Central Research Question & Purpose

• Investigate how three kinds of teacher feedback (ability, effort, negative) relate to pupils’ self-talk (positive & negative) and academic self-concept (evaluative & descriptive) within two specific domains – reading and mathematics.
• Test Burnett’s (1999) mediating model that positions self-talk as the mechanism through which feedback influences self-concept (subject-specific extension of Craven, Marsh & Debus, 1991, and Blote, 1995).

Conceptual & Theoretical Background

1. Blote’s (1995) Teacher-Expectation Model

• Steps: (A) teacher expectations → (B) differential feedback → (C) student internalisation (self-talk) → (D) aligned student self-expectations (self-concept).

2. Craven, Marsh & Debus’ (1991) Internal Mediating Process Model

• (A) Teacher gives performance feedback → (B) Student internalises (“I did well”) → (C) Generalises to subject (“I’m good at reading”).

3. Attributional Feedback Theory

• Ability feedback: Attributes success to stable ability (e.g., “You’re really smart at reading”).
• Effort feedback: Attributes success to controllable effort (e.g., “You tried really hard”).
• Empirical tension:
  • Marsh (1990), Craven et al. (1991) – ability feedback boosts self-concept.
  • Mueller & Dweck (1998) – exclusive ability praise harms motivation after failure; effort praise promotes persistence.

Hypothesised Subject-Specific Model (Figure 3)

Teacher Feedback  →  Self-Talk  →  Evaluative Self-Concept  →  Descriptive Self-Concept
      (ability, effort,            (positive, negative)          (competence judgement)        (liking/enjoyment)
       negative)

Instruments & Reliability

Teacher Feedback Scale (TFS)

• 24 items (reading & maths versions).
• 3 sub-scales, each scored $3=\text{Often},\;2=\text{Sometimes},\;1=\text{Never}$
  • Ability (αread = 0.84, αmath = 0.88)
  • Effort (αread = 0.83, αmath = 0.83)
  • Negative (αread = 0.80, αmath = 0.78)

Self-Talk Inventory (STI)

• 20 context-specific statements per subject (10 reading, 10 maths).
• Positive self-talk (αread = 0.73; αmath = 0.67)
• Negative self-talk (αread = 0.62; αmath = 0.77)
• Nominal response: Often / Sometimes / Never.

The Self Scale (TSS)

• 8 items per subject.
• Two latent facets:
  • Evaluative (“I am good at maths”) – αread = 0.81; αmath = 0.85
  • Descriptive (“I like reading”) – αread = 0.87; αmath = 0.89

Sample Item Sets

• Reading positive: “I can do this.” / “I am a good reader.”
• Reading negative: “Why me?” / “I am hopeless at reading in front of a group.”
• Maths positive: “I can do this.” / “I am good at maths.”
• Maths negative: “There is no way I can do this.” / “Maths has always been a problem for me.”

Participants & Context

• Total $n = 747$ (Grades 3-6). 396 boys (53 %), 351 girls (47 %).
• Mean age $\bar{x} = 9.9\;\text{yrs},\; SD = 1.2\;\text{yrs}.\;\text{Range}: 7–12.$
• Socio-economic context: lower-middle-class; predominantly European background.

Procedure

1. Research assistant administered TFS, STI, TSS in regular class sessions (25-30 students).
2. Items read aloud for any student needing assistance.

Data Analysis – Structural Equation Modelling (SEM)

• Software: LISREL 7.0 within SPSS.
• Fit indices adopted (desired >.90 for good fit): $\chi^2$, GFI, AGFI, Tucker-Lewis Index (TLI), Relative Non-centrality Index (RNI), RMSR.

Results

Reading Model (Figure 4)

• Fit: \chi^2(62)=132,\; p<.001;\; GFI=.98;\; AGFI=.96;\; TLI=.98;\; RNI=.98;\; RMSR=.04.
• Significant (standardised) paths:
  • Ability → Positive ST $=.57$
  • Ability → Negative ST (+)
  • Effort → Negative ST $=.30$ (paradoxical)
  • Negative FB → Negative ST $=.67$
  • Positive ST → Evaluative SC $=.62$
  • Negative ST → Evaluative SC $=-.27$
  • Negative ST → Positive ST $=-.15$
  • Evaluative SC → Descriptive SC $=.67$
• Variance explained:
  • Positive ST ≈ 32 %; Negative ST ≈ 45 %; Evaluative SC ≈ 50 %; Descriptive SC ≈ 45 %.

Mathematics Model (Figure 5)

• Fit: \chi^2(62)=222,\; p<.001;\; GFI=.96;\; AGFI=.93;\; TLI=.96;\; RNI=.97;\; RMSR=.05.
• Paths largely mirror reading model with two notable differences:
  • Ability → Negative ST stronger (≈ double magnitude).
  • Positive ST → Descriptive SC weak but significant $=.13$.

Key Patterns Across Both Domains

• Mediation confirmed: Self-talk fully/partially carries the effect of feedback to self-concept.
• Ability feedback – simultaneously boosts positive self-talk and (unexpectedly) some negative self-talk.
• Effort feedback – unexpectedly raises negative self-talk (students may interpret effort praise as “I lack ability”).
• Negative teacher comments strongly heighten negative self-talk; no direct path to self-concept.
• Students’ competency judgments (evaluative SC) strongly predict enjoyment/liking (descriptive SC).

Discussion & Interpretation

• Empirical support for Burnett’s (1999) mediating framework in subject-specific settings.
• Confirms critical role of self-talk as the cognitive mechanism linking external teacher input to internal self-beliefs.
• Counter-intuitive finding: Effort praise correlates with greater negative self-talk – may reflect Australian pupils’ preference hierarchy (Burnett 2001) or implicit message “you need to try because you’re not good.”
• Ability feedback remains a double-edged sword: promotes positive self-talk/self-concept but increases negative rumination when ability attributions are interpreted rigidly.

Limitations

• Teacher feedback measured via student perception (no observational corroboration).
• Cross-sectional design; causality inferred from SEM, not manipulated.
• Reliability of negative self-talk in reading modest (α = 0.62).

Practical Implications for Educators

• Balance feedback: Combine ability & effort praise; avoid exclusive dependence on either.
• Monitor interpretation: Discuss with students how they perceive different kinds of praise.
• Minimise negative statements: Direct link from negative feedback → negative self-talk → ↓ self-concept.
• Scaffold failure episodes: After errors, frame feedback to protect self-concept (e.g., strategy-based coaching).

Connections to Broader Literature

• Aligns with expectancy-value frameworks (Weinstein et al., 1987) and mindset research (Dweck).
• Supports Marsh’s internal/external frame-of-reference notion: pupils integrate feedback with self-comparisons.

Selected Numerical & Statistical Highlights

• Sample: $n = 747$; Grades 3–6.
• Reliability range: $\alpha = .62\text{ to } .89$ across all sub-scales.
• Fit indices all >.95 (except AGFI for maths =.93) → very good model fit.

Reference Snapshot (cited in transcript)

• Babad (1990a, 1990b) – differential teacher behaviour.
• Blote (1995) – teacher expectation model.
• Blumenfeld et al. (1982); Brattesani et al. (1984); Cooper & Good (1983); Parsons et al. (1982); Weinstein et al. (1987).
• Burnett series (1994 → 2001) on self-talk, feedback, self-concept.
• Craven, Marsh & Debus (1991) – attributional feedback.
• Mueller & Dweck (1998) – praise effects.
• Marsh (1990) – internal vs external reference.