Chapter 2:

Understanding Data in
Second Language Acquisition
(SLA) Research
Essential Methods for Studying How We
Learn New Languages

WEEK 03
Learning objectives:
Chapter 2. Where Do Data Come From? (GT1: Second language
acquisition: An introductory course)
I. Data Types
II. Learner Corpora
III. Data elicitation
IV. Replication
V. Meta-Analyses
VI. Issues in Data Analysis
[Link] is Acquisition?
 Why is Data Important in SLA Research?

Defining SLA Data Interdisciplinary Driving

Roots Improvement

Information collected to Ideas and methods borrowed from Improve teaching methods
diverse academic fields.
understand how people learn Create better learning materials
Linguistics
a second language. Understand learner growth
Child Language Acquisition
process
Sociology
Clues for Language Learning
Psychology

Deep Dive: Key Data Approaches

Learner Corpora Longitudinal Insights

Systematic collections of authentic, continuous language use Yield most valuable insights into second language
from L2 learners. acquisition over time.
Authentic Data Sources Developmental Patterns
 Groupwork: Mind-mapping
Task: On a piece of paper, write two key terms: longitudinal studies
and cross-sectional studies.
Work in groups, read the material given again, and write down key
ideas for the following categories:
❖ Definitions + Example
❖ Core characteristics
❖ Major advantages
❖ Challenges/ Limitations
 Two Main Ways to Collect Data: Longitudinal vs.
Cross-Sectional
Longitudinal Studies: Cross-Sectional Studies: A
Watching Growth Over Time Snapshot in Time

"Like observing a plant grow day by day . "Like taking a single photo of many different
Focuses on how development unfolds." plants at various growth stages . Focuses on
what exists at a given moment."
• Process-Oriented • Captures Change • Broad Overview • Comparative Analysis

• Developmental Insights • Time-Specific

Research Question is Key!

SLA Insights: Value of Longitudinal Data
Your research question usually guides which method is best
"Longitudinal data can potentially yield the most valuable insights into
for your study. second-language acquisition."
"Second language acquisition is highly systematic."
Longitudinal Studies:
Following the Language Journey
Core Characteristics Key Advantages
Case studies (often single subject) Shows developmental trends
Extended period (weeks to years) Tracks individual learning paths
Regular data collection Reveals input & social factors

Primary Data Type Challenges

Spontaneous speech Time-consuming collection & analysis

Rich, natural language Limited generalizability

Conceptual Language Proficiency Over

Time
Kumpf (1984) studied Uguisu, a 5-year-old Japanese girl
learning English. Analyzing her spontaneous speech
revealed how her language system developed naturally,
especially temporality and verb tenses.
Cross-Sectional Studies: Capturing a Moment
Key Characteristics
Key Advantages
(a) Number of subjects (Data: large numbers of participants) &
Timeframe: Collected at a single point in time. Greater generalizability (large samples).
➢ Uses controlled tasks or tests. Much quicker to conduct.
(b) Data Type & Focus Focus on quantitative analysis.
Snapshot Moment in Time
➢ Elicits specific language features.
➢ Tests specific research hypotheses.

Key Characteristics Limitations

(c) Descriptive detail Provides lessBroad
Efficiency individual
Reach detail.
Doesn't show *how* learning develops.
(d) Analysis of data: Captures a 'slice' of development.
Reveals only the *current state*.
Controlled Data Hypothesis-Driven Limited Depth Process Gap

Example: Gass & Ard (1984) Pseudolongitudinal Studies.

Investigated learner understanding of the English Compares different proficiency groups at once
progressive (e.g., 'I am eating'). They used tasks Assumes groups represent individual growth stages
like sentence judging, grouping, and writing to This assumption can be controversial.
see which meanings were easier for a large group
of learners at one time, creating a snapshot of
their ability.
 Longitudinal studies >< Cross-sectional studies
Fill in the table with the information from the given material.
Longitudinal studies >< Cross-sectional studies
 Longitudinal studies >< Cross-sectional studies
Which examples belong to: a) Longitudinal studies b) Cross-sectional studies

A study compares beginner, intermediate, and advanced English learners from Spain to see
how each group uses articles (a, an, the). Beginners may omit articles, intermediates may
overuse the, and advanced learners approximate native-like use. b

A researcher records the oral production of 20 Vietnamese students learning English over
three years. Every six months, their spoken English is analyzed to see how past tense forms
(e.g., walked, went) emerge and stabilize. Findings might show that learners first overuse
irregular forms (goed instead of went), but gradually replace them with correct forms. a

Comparing beginners, intermediates, and advanced French learners of English in one year
shows progression in using prepositions (in, on, at). b

Following 10 Japanese learners in Canada for 5 years reveals how they gradually acquire
English question forms (from You like coffee? → Do you like coffee?). a
Longitudinal studies >< Cross-sectional studies
Cross-Sectional Studies: Capturing a Moment
Learner Corpora: Big Data for Language Learning
What Are Learner Corpora? Key Advantages
Huge, organized collections of L2 learner data.
Automated analysis of vast data.
Massive digital libraries: essays, speeches, etc.
Provides strong statistical evidence.
L2 Learner Data Digital Libraries Authentic Language
Insights into naturalistic language use.
Many are freely available online (e.g., ICLE).

Statistical Evidence Longitudinal Insights Open Access

Analysis Methods Potential Disadvantages

Use programs (concordancers) to
No control over how existing data was collected.
find patterns.
Participant demographics might be unknown or biased.
Analyze errors, frequency, and
Can limit the specific research questions you can ask.
collocations.
Computer Analysis

Pattern Recognition Data Collection Bias Limited Control Research Scope

Main Steps in Learner Corpus Research

1. Select/Create Corpus 2. Annotate Data 3. Extract Data 4. Analyze & Interpret

Page 27
Page 28
 Eliciting Language: Making Learners Produce Data 7
Measuring General Proficiency Verbal Report Data
How good is someone at a language?
What are learners thinking?

Standardized Tests (TOEFL, IELTS, CEFR) Think-Alouds: Real-time cognitive processes

Informal Measures (Classroom, immersion) WhatStimulated Recall
are learners ⚫ Post-Production Interviews
thinking?

Language Elicitation Measures Narrative Inquiry

Getting specific language samples. Collecting stories about learning journeys.
Elicited Imitation Grammaticality Judgments Identity Motivation Personal Experiences

DCQs for Pragmatics

(Discourse Completion Questionnaires

Reference: From second language acquisition research to foreign language ... | ScienceDirect
Data elicitation: Measuring general proficiency

(a) impressionistic judgments

(b) institutional status
(c) specific research-designed test
(d) standardized tests
 Measuring general proficiency
• Impressionistic judgments – based on a
teacher’s or researcher’s overall sense of a
learner’s ability.
• Institutional status (“seat time”) – determined
by how long or at what level a learner has
studied (e.g., second-year French).
• Tests – either specially designed for a study or
standardized (like TOEFL, CEFR, ACTFL, ILR).
• Self-ratings – learners evaluate their own skills
(reading, writing, speaking), though reliability
depends on careful design of the tool.
 Verbal Report Data

1/ Think-alouds

2/ Stimulated recalls

3/ Post-production interviews
1. Think-Alouds

❖ Learners verbalize their thoughts during a task (e.g., reading or

writing).
❖ Researchers can observe strategies learners use (guessing word
meanings, checking grammar, reformulating sentences).
❖ Useful for reading and writing tasks, but not feasible for oral tasks
(since one can’t talk about thoughts while speaking).
 2. Stimulated Recalls
❖ Learners recall their thought processes after a
task.
❖ Requires a stimulus (e.g., video, audio, or written
work) to jog memory.
❖ Especially useful for oral communication research.
❖ Risk: recalls may drift into present-time reflections
instead of past thoughts if questions are poorly
designed.
❖ Best practice: ask neutral, task-focused questions
(e.g., “What were you thinking when you said…?”).
•.

3. Post-Production Interviews
❖ Conducted after tasks are
completed, without relying on
immediate recall.
❖ Aimed at gathering learners’
reflections, attitudes, or
preferences about tasks or activities
(e.g., which was easiest, most useful,
most familiar).
❖ Complements empirical data by
providing insights into learner
perceptions.
 Exploring Cognitive Processes: How the Brain Learns

Response Speed Processing Difficulty

Measures how quickly someone responds to a Learners read sentences word by word,
stimulus. controlling the pace.
Faster reaction times often mean more automatic Longer time on a word can indicate processing difficulty
processing (e.g., pressing a button for an English word). (e.g., unfamiliar vocabulary or grammar).

Neurolinguistic Data: Inside the Brain!

Attention & Effort
Event-Related Potentials (ERPs)
Attention & Effort Measures millisecond electrical brain
Monitors where a learner's eyes look on a Event-Related
activity. (N400 forPotentials (ERPs)
surprise, P600 for
screen or page. grammar errors).

Functional MRI (fMRI)

Reveals what words they focus on, indicating attention and
processing effort. Measures blood flow to see which brain
Functional
areas are activeMRI (fMRI)
during language tasks.
 Key Challenges in SLA Data Analysis
1. Measuring 2. Interpreting 3. Defining
Development Results 'Acquisition'
How do we truly know if a Learners have their own unique language When is a feature truly 'acquired'? It's not
system ('interlanguage'), different from the first correct use, as learners can
learner is improving? It’s not as
their native (NL) and target languages backslide' and lose accuracy.
simple as in child language (TL).
(e.g., MLU).

Interlanguage System
T-unit: Syntactic Complexity Backsliding is common
Comparative Fallacy to Avoid
C-unit/AS-unit: Oral Production Criteria: 90% accuracy
Example: Hmong speaker's
Error-free units for L2 learners articles Focus on 'Emergence' over time

"Longitudinal data yields the most valuable insights into

second-language acquisition processes."

- SLA Research Consensus

Source: SLA Week [Link] (Section 2.6, 2.7)

 KEY POINTS TO REMEMBER
❑ Think-alouds: performed during the task >< stimulated recall: performed after the task
=> Both require the learner to reflect on his/her performance.

❑ Stimulated recall: a stimulus that stimulates the learner to recall his/her thought
processes at the time of performing the task.

❑ Intuitional data: come from participants being asked about their intuitions about
whether a given sentence is grammatical in a certain language

❑ Reaction time data: a measurement of how long it takes someone to respond to a

particular stimulus.

❑ T-units and C-units: measures of syntactic development common in L2 research.

 KEY POINTS TO REMEMBER
❑ An AS-unit: a measure of development commonly used for oral
production.

❑ Backsliding: instances when correct forms appear but then seem to

disappear in a learner’s IL

❑ Language games: elicit data while removing the focus from the language
or particular language structures.

❑ Discourse completion questionnaires: gather data about pragmatics, or

how the language is used, including such speech acts as apologizing,
refusing, or requesting => ask learners and native speakers about what
they would say in certain situations.
Why Replication and Meta-Analysis
Matter
Essential Tools for Robust SLA Research
Replication: Can we get the Meta-Analysis: Combining
same results again? studies for a bigger picture!
Repeating studies with varied conditions
Statistical synthesis of multiple studies
Crucial for verifying results in SLA
Ensures robustness & generalizability Uses effect sizes to find trends
Objective, transparent, powerful evidence
SLA involves complex human behavior & individual
Identifies stronger effects, building robust theories &
differences, making replication vital.
practices in SLA.

Reproducibility Validation SLA Foundations Synthesis Effect Sizes Evidence-Based

References:
Grit and L2 grit research in SLA (2013–2025) | Cambridge Core
From second language acquisition research to foreign language ... | ScienceDirect
Learner Corpus Research and Second Language Acquisition | Cambridge Core
Conclusion: The Rich World of SLA Research Methods 11

SLA research employs a wide array of methods

Choosing the right method is guided by the specific research
to understand how individuals learn new question and the insights you aim to uncover. Each offers
languages. unique advantages.

Analyzing Brain Activity Understanding methodologies is crucial for:

Observing Natural Speech
Critically evaluating research findings
Learner Corpora Longitudinal Studies
Developing effective teaching strategies

Enhancing language learning

Empowering English Majors

Recognize research tools to enhance linguistics

understanding, engage, and contribute to the field of SLA.
Keep exploring!

References:
Learner Corpus Research and Second Language Acquisition; The potential of publicly shared longitudinal learner corpora in SLA...
 Matching Exercise: SLA Research Method
Instructions: Match each research method (Column A) with its correct
definition or main characteristic (Column B). Write the correct letter (A–J) next to
each term.
Column A – Terms
1. Longitudinal studies
2. Cross-sectional studies
3. Learner corpora
4. Standardized language tests
5. Questionnaires (psychology tests)
6. Verbal report data (think-alouds, stimulated recall, post-production
interviews)
7. Language elicitation measures
8. Intuitional data
9. Language games
10. Discourse completion questionnaires (DCQs)
 Matching Exercise: SLA Research Method
Column B – Definitions / Characteristics
A. Large, systematic collections of written or spoken texts by learners, analyzed for
errors, patterns, and vocabulary use.
B. Tasks or prompts designed to elicit specific language forms (e.g., describing a picture
to study tense use).
C. Same group of learners studied repeatedly over a long period to observe
developmental changes.
D. Short tasks or role-play activities designed to produce spontaneous language in a
playful, low-pressure way.
E. Structured instruments (like surveys) used to collect learner attitudes, beliefs, or
psychological traits such as motivation or anxiety.
 Matching Exercise: SLA Research Method
Column B – Definitions / Characteristics
F. Learners’ reflections or explanations of their thought processes while performing or
recalling a task.
G. Hypothetical scenarios that ask learners to produce what they would say in certain
social or pragmatic contexts.
H. Research that compares different groups of learners at one point in time to infer
developmental stages.
I. Introspective judgments by native speakers or learners about whether certain
sentences are grammatical/acceptable.
J. Widely used, norm-referenced tests (e.g., TOEFL, IELTS) that provide comparable
scores across learners.
 Question & Answer
Q1. What is the main feature of a longitudinal study in SLA research?
a) It collects data from many learners at the same time.
b) It follows the same learners over an extended period.
c) It compares native and non-native speakers on one test.
d) It analyzes a learner’s intuition about grammar.
Q2. What is cross-sectional study?
Q3. What is a learner corpora? Provide one practical example of how it can be used in
language teaching.
Q4. Why are standardized tests (e.g., TOEFL, IELTS) important in SLA research?
a) They are only used to diagnose psychological problems.
b) They allow researchers to compare learners across contexts using the same criteria.
c) They focus only on speaking fluency.
d) They replace classroom-based assessments
 Question & Answer
Q5. What kind of learner variables can psychology-oriented questionnaires
measure? Give two examples
Q6. Which of the following is NOT a type of verbal report data?
a) Think-alouds
b) Stimulated recall
c) Post-production interviews
d) Multiple-choice grammar tests
Q7. Give one classroom task that could serve as a language elicitation measure
to study learners’ grammar use.
 Question & Answer
Q8. In SLA research, intuitional data usually comes from:
a) Learners’ written essays
b) Teachers’ lesson plans
c) Judgments about sentence grammaticality
d) TOEFL reading passages
Q9. How can language games be used in SLA research? Provide one example.
Q10. What do discourse completion questionnaires (DCQs) typically ask
learners to do?
a) Recall grammar rules from memory
b) Produce what they would say in hypothetical social situations
c) Translate sentences from L1 to L2
d) Take a standardized multiple-choice test