**Introduction to Language Testing**

**Comprehensive Lecture Notes**

(Updated as of January 2026 – reflecting key developments through late 2025)

### 1. Definition and Core Purpose

**Language testing / language assessment** is the systematic process of gathering evidence about a
person's language ability to support decision-making.

Core decisions include:

- University/program admission (proficiency)

- Class level placement

- Course achievement/certification

- Identification of specific learning needs (diagnostic)

- Program evaluation & policy

Contemporary shift: From assessment **of** learning → assessment **for/as** learning (formative,
learning-oriented)

### 2. Main Purposes of Language Tests (2026 Classification)

| Purpose | Primary Decision | Main Stakeholders |

Representative Examples (2026) |

|--------------------------------|-----------------------------------------------|------------------------------------|-----------------
--------------------------------|

| Proficiency | General readiness for academic/professional use | Universities, employers,

immigration | IELTS, TOEFL iBT, PTE Academic, Duolingo English Test |

| Achievement | Mastery of taught syllabus | Teachers, schools, ministries |

End-of-term exams, national school-leaving tests |

| Diagnostic | Pinpoint strengths/weaknesses | Teachers, learners, tutors |

Formative tools, AI-powered diagnostic apps |
| Placement | Appropriate course/level assignment | Language centers, universities
| Institutional adaptive placement tests |

| Learning-Oriented Assessment (LOA) | Support & enhance ongoing learning | Teachers +

students | AI feedback loops, dynamic assessment, portfolios |

### 3. Historical Evolution (Key Paradigm Shifts)

| Period | Dominant Paradigm | Characteristics | Landmark

Tests/Developments |

|----------------------------|---------------------------------------|--------------------------------------------------|---------------
----------------------------------|

| Pre-1950s | Grammar-translation | Translation, rules, literary texts |

Traditional school/university exams |

| 1950s–1970s | Structural / discrete-point | Isolated items (phonology, grammar, vocab)

| Early TOEFL paper-based, Lado tests |

| 1970s–mid-1980s | Integrative / pragmatic | Cloze, dictation, essay, interview |

TOEFL 1976 version, early Cambridge exams |

| Late 1980s–2000s | Communicative competence | Task-based, real-life, performance

| IELTS (1989), ACTFL OPI, CAE/CPE |

| 2010s–early 2020s | Technology + post-communicative | Computer-adaptive, automated

scoring | PTE Academic (fully auto), TOEFL iBT, Duolingo |

| 2023–2026 | AI / GenAI era + ethical focus | Multimodal, adaptive, embedded in learning

| Advanced DET, AI-augmented Cambridge/ETS tools |

### 4. Core Quality Criteria (Still Centered on Bachman & Palmer + Contemporary Extensions)

| Criterion | Core Meaning | 2026 Emphasis & Challenges |

|------------------------|---------------------------------------------------|--------------------------------------------------------|

| **Reliability** | Score consistency (test–retest, inter-rater) | AI-human hybrid reliability;

consistency across models |

| **Construct Validity** | Measures intended language ability | Unitary view + consequences

(Messick → Kane → AI era) |
| **Authenticity** | Resemblance to real-world use | Multimodal tasks, translanguaging,
online interaction |

| **Interactiveness** | Engagement of learner resources & strategies | Adaptive AI increases

personalization |

| **Impact / Washback** | Effects on teaching, learning, society | Positive LOA vs. negative test-
prep pressure |

| **Practicality** | Resources, time, cost, scalability | AI dramatically improves (speed, cost,

access) |

| **Fairness** | Equity, bias absence, accessibility | Major priority: AI bias mitigation,

digital divide |

### 5. Dominant Theoretical Frameworks (2026)

- **Bachman & Palmer (2010)** → Organizational + Pragmatic competence (still foundational)

- **CEFR Companion Volume (2020)** → Mediation, online interaction, plurilingual/pluricultural

competence, sign language descriptors (no major structural update post-2020, but widely implemented)

- Socio-cognitive framework (Weir & others) → Digital contexts

- Argument-based validation (Kane, Chapelle) → Increasingly applied to AI scoring

### 6. Major Contemporary Trends (2025–2026 Snapshot)

1. **AI & Generative AI Explosion**

- Automated scoring (writing/speaking) → high human correlation but debate on higher-order

constructs

- Adaptive, real-time, data-rich assessment embedded in learning (moving "beyond testing")

- Tools: ChatGPT-based custom scorers, advanced SpeechRater, ELSA Speak, Duolingo AI upgrades

2. **Learning-Oriented & Formative Focus**

- Assessment → support learning (immediate feedback, personalized analytics)

- Hybrid AI-human models recommended for balance

3. **Fairness, Ethics & Bias Mitigation**

- Bias in training data → ongoing concern

- Frameworks for transparency, explainability, equity

- Human oversight for high-stakes decisions

4. **Multimodal & Inclusive Assessment**

- Visuals + gestures + digital literacy

- Growing acceptance of plurilingual resources

### 7. Quick Comparison of Major International Tests (Early 2026)

| Test | Format | Scale | Auto-scoring | Typical Duration | Approx. Cost |

Strongest Strength | Main Criticism / Limitation |

|------------------------|-------------------------------|----------------|-------------------|------------------|--------------|------
----------------------------------|----------------------------------------|

| IELTS Academic | Paper/computer + human speaking | 0–9 bands | Partial (pilots) | ~2h45m
| High | Immigration + global recognition | Rater variability, longer wait |

| TOEFL iBT | Computer (some adaptive) | 0–120 | High (Speaking/Writing) | ~2h |

High | Strong academic orientation | Cost, test length |

| PTE Academic | Fully computer | 10–90 | Almost complete | ~2h | Medium

| Fast results (usually 48h) | Perceived as "mechanical" |

| Duolingo English Test | Fully online, adaptive | 10–160 | Complete | ~1h | Very
low | Convenience, accessibility, low cost | Lower acceptance at elite institutions |

### 8. Practical Checklist for Teachers / Test Users (2026 Perspective)

When designing, selecting or evaluating a test:

- Is the **construct** clearly defined & current (includes mediation/online/plurilingual aspects)?

- Is there solid evidence of **reliability** (human + AI components)?

- Does it support **positive washback** & learning-oriented outcomes?

- How **authentic** & **interactive** are the tasks?

- Is it **fair** & equitable (bias checks, accessibility, digital divide)?

- Is **practicality** acceptable (cost, time, tech requirements)?

- Does it provide **actionable, timely feedback**?

These notes are suitable for a 90–120 minute university lecture. The field is currently dominated by the
rapid integration of AI, ethical concerns, and the move toward assessment embedded in learning rather
than standalone testing. Feel free to add local Nepali examples (e.g., university entrance tests, SEE
English assessment) for greater relevance in Kathmandu classrooms.